Skip to content

Instantly share code, notes, and snippets.

@stas-sl
Last active June 8, 2016 23:52
Show Gist options
  • Save stas-sl/b6460d3f5c66c8d2bb6b27737b317868 to your computer and use it in GitHub Desktop.
Save stas-sl/b6460d3f5c66c8d2bb6b27737b317868 to your computer and use it in GitHub Desktop.
This file has been truncated, but you can view the full file.
I0607 01:01:32.950039 32403 solver.cpp:280] Solving mixed_lstm
I0607 01:01:32.950052 32403 solver.cpp:281] Learning Rate Policy: fixed
I0607 01:01:32.960855 32403 solver.cpp:338] Iteration 0, Testing net (#0)
I0607 01:02:33.209952 32403 solver.cpp:393] Test loss: 2.53488
I0607 01:02:33.210446 32403 solver.cpp:406] Test net output #0: loss1/accuracy = 0.638494
I0607 01:02:33.210469 32403 solver.cpp:406] Test net output #1: loss1/accuracy01 = 0.796
I0607 01:02:33.210481 32403 solver.cpp:406] Test net output #2: loss1/accuracy02 = 0.667
I0607 01:02:33.210494 32403 solver.cpp:406] Test net output #3: loss1/accuracy03 = 0.569
I0607 01:02:33.210505 32403 solver.cpp:406] Test net output #4: loss1/accuracy04 = 0.476
I0607 01:02:33.210517 32403 solver.cpp:406] Test net output #5: loss1/accuracy05 = 0.534
I0607 01:02:33.210530 32403 solver.cpp:406] Test net output #6: loss1/accuracy06 = 0.732
I0607 01:02:33.210541 32403 solver.cpp:406] Test net output #7: loss1/accuracy07 = 0.859
I0607 01:02:33.210552 32403 solver.cpp:406] Test net output #8: loss1/accuracy08 = 0.923
I0607 01:02:33.210564 32403 solver.cpp:406] Test net output #9: loss1/accuracy09 = 0.968
I0607 01:02:33.210577 32403 solver.cpp:406] Test net output #10: loss1/accuracy10 = 0.985
I0607 01:02:33.210588 32403 solver.cpp:406] Test net output #11: loss1/accuracy11 = 0.996
I0607 01:02:33.210600 32403 solver.cpp:406] Test net output #12: loss1/accuracy12 = 0.999
I0607 01:02:33.210611 32403 solver.cpp:406] Test net output #13: loss1/accuracy13 = 1
I0607 01:02:33.210623 32403 solver.cpp:406] Test net output #14: loss1/accuracy14 = 1
I0607 01:02:33.210634 32403 solver.cpp:406] Test net output #15: loss1/accuracy15 = 1
I0607 01:02:33.210646 32403 solver.cpp:406] Test net output #16: loss1/accuracy16 = 1
I0607 01:02:33.210657 32403 solver.cpp:406] Test net output #17: loss1/accuracy17 = 1
I0607 01:02:33.210669 32403 solver.cpp:406] Test net output #18: loss1/accuracy18 = 1
I0607 01:02:33.210680 32403 solver.cpp:406] Test net output #19: loss1/accuracy19 = 1
I0607 01:02:33.210691 32403 solver.cpp:406] Test net output #20: loss1/accuracy20 = 1
I0607 01:02:33.210703 32403 solver.cpp:406] Test net output #21: loss1/accuracy21 = 1
I0607 01:02:33.210714 32403 solver.cpp:406] Test net output #22: loss1/accuracy22 = 1
I0607 01:02:33.210726 32403 solver.cpp:406] Test net output #23: loss1/accuracy_incl_empty = 0.894457
I0607 01:02:33.210737 32403 solver.cpp:406] Test net output #24: loss1/accuracy_top3 = 0.851189
I0607 01:02:33.210754 32403 solver.cpp:406] Test net output #25: loss1/cross_entropy_loss = 1.36357 (* 0.3 = 0.40907 loss)
I0607 01:02:33.210770 32403 solver.cpp:406] Test net output #26: loss1/cross_entropy_loss_incl_empty = 0.395468 (* 0.3 = 0.11864 loss)
I0607 01:02:33.210784 32403 solver.cpp:406] Test net output #27: loss1/loss01 = 0.880903 (* 0.0272727 = 0.0240246 loss)
I0607 01:02:33.210798 32403 solver.cpp:406] Test net output #28: loss1/loss02 = 1.31433 (* 0.0272727 = 0.0358454 loss)
I0607 01:02:33.210811 32403 solver.cpp:406] Test net output #29: loss1/loss03 = 1.59453 (* 0.0272727 = 0.0434871 loss)
I0607 01:02:33.210825 32403 solver.cpp:406] Test net output #30: loss1/loss04 = 1.79761 (* 0.0272727 = 0.0490257 loss)
I0607 01:02:33.210839 32403 solver.cpp:406] Test net output #31: loss1/loss05 = 1.53639 (* 0.0272727 = 0.0419016 loss)
I0607 01:02:33.210852 32403 solver.cpp:406] Test net output #32: loss1/loss06 = 0.965465 (* 0.0272727 = 0.0263309 loss)
I0607 01:02:33.210865 32403 solver.cpp:406] Test net output #33: loss1/loss07 = 0.542771 (* 0.0272727 = 0.0148029 loss)
I0607 01:02:33.210882 32403 solver.cpp:406] Test net output #34: loss1/loss08 = 0.284592 (* 0.0272727 = 0.00776159 loss)
I0607 01:02:33.210896 32403 solver.cpp:406] Test net output #35: loss1/loss09 = 0.158202 (* 0.0272727 = 0.00431459 loss)
I0607 01:02:33.210911 32403 solver.cpp:406] Test net output #36: loss1/loss10 = 0.075429 (* 0.0272727 = 0.00205715 loss)
I0607 01:02:33.210924 32403 solver.cpp:406] Test net output #37: loss1/loss11 = 0.0244008 (* 0.0272727 = 0.000665477 loss)
I0607 01:02:33.210938 32403 solver.cpp:406] Test net output #38: loss1/loss12 = 0.0152086 (* 0.0272727 = 0.000414779 loss)
I0607 01:02:33.210952 32403 solver.cpp:406] Test net output #39: loss1/loss13 = 0.0101094 (* 0.0272727 = 0.000275711 loss)
I0607 01:02:33.210981 32403 solver.cpp:406] Test net output #40: loss1/loss14 = 0.00725043 (* 0.0272727 = 0.000197739 loss)
I0607 01:02:33.210996 32403 solver.cpp:406] Test net output #41: loss1/loss15 = 0.00511528 (* 0.0272727 = 0.000139508 loss)
I0607 01:02:33.211010 32403 solver.cpp:406] Test net output #42: loss1/loss16 = 0.00388001 (* 0.0272727 = 0.000105819 loss)
I0607 01:02:33.211024 32403 solver.cpp:406] Test net output #43: loss1/loss17 = 0.00343643 (* 0.0272727 = 9.37208e-05 loss)
I0607 01:02:33.211038 32403 solver.cpp:406] Test net output #44: loss1/loss18 = 0.00300968 (* 0.0272727 = 8.20821e-05 loss)
I0607 01:02:33.211051 32403 solver.cpp:406] Test net output #45: loss1/loss19 = 0.00226606 (* 0.0272727 = 6.18016e-05 loss)
I0607 01:02:33.211066 32403 solver.cpp:406] Test net output #46: loss1/loss20 = 0.002128 (* 0.0272727 = 5.80364e-05 loss)
I0607 01:02:33.211079 32403 solver.cpp:406] Test net output #47: loss1/loss21 = 0.00147219 (* 0.0272727 = 4.01505e-05 loss)
I0607 01:02:33.211093 32403 solver.cpp:406] Test net output #48: loss1/loss22 = 0.000934915 (* 0.0272727 = 2.54977e-05 loss)
I0607 01:02:33.211105 32403 solver.cpp:406] Test net output #49: loss2/accuracy = 0.796195
I0607 01:02:33.211117 32403 solver.cpp:406] Test net output #50: loss2/accuracy01 = 0.89
I0607 01:02:33.211129 32403 solver.cpp:406] Test net output #51: loss2/accuracy02 = 0.841
I0607 01:02:33.211140 32403 solver.cpp:406] Test net output #52: loss2/accuracy03 = 0.792
I0607 01:02:33.211153 32403 solver.cpp:406] Test net output #53: loss2/accuracy04 = 0.677
I0607 01:02:33.211163 32403 solver.cpp:406] Test net output #54: loss2/accuracy05 = 0.698
I0607 01:02:33.211174 32403 solver.cpp:406] Test net output #55: loss2/accuracy06 = 0.798
I0607 01:02:33.211185 32403 solver.cpp:406] Test net output #56: loss2/accuracy07 = 0.892
I0607 01:02:33.211197 32403 solver.cpp:406] Test net output #57: loss2/accuracy08 = 0.937
I0607 01:02:33.211208 32403 solver.cpp:406] Test net output #58: loss2/accuracy09 = 0.969
I0607 01:02:33.211220 32403 solver.cpp:406] Test net output #59: loss2/accuracy10 = 0.984
I0607 01:02:33.211231 32403 solver.cpp:406] Test net output #60: loss2/accuracy11 = 0.995
I0607 01:02:33.211242 32403 solver.cpp:406] Test net output #61: loss2/accuracy12 = 0.998
I0607 01:02:33.211253 32403 solver.cpp:406] Test net output #62: loss2/accuracy13 = 1
I0607 01:02:33.211264 32403 solver.cpp:406] Test net output #63: loss2/accuracy14 = 1
I0607 01:02:33.211275 32403 solver.cpp:406] Test net output #64: loss2/accuracy15 = 1
I0607 01:02:33.211287 32403 solver.cpp:406] Test net output #65: loss2/accuracy16 = 1
I0607 01:02:33.211297 32403 solver.cpp:406] Test net output #66: loss2/accuracy17 = 1
I0607 01:02:33.211308 32403 solver.cpp:406] Test net output #67: loss2/accuracy18 = 1
I0607 01:02:33.211319 32403 solver.cpp:406] Test net output #68: loss2/accuracy19 = 1
I0607 01:02:33.211330 32403 solver.cpp:406] Test net output #69: loss2/accuracy20 = 1
I0607 01:02:33.211341 32403 solver.cpp:406] Test net output #70: loss2/accuracy21 = 1
I0607 01:02:33.211352 32403 solver.cpp:406] Test net output #71: loss2/accuracy22 = 1
I0607 01:02:33.211364 32403 solver.cpp:406] Test net output #72: loss2/accuracy_incl_empty = 0.938501
I0607 01:02:33.211375 32403 solver.cpp:406] Test net output #73: loss2/accuracy_top3 = 0.920895
I0607 01:02:33.211388 32403 solver.cpp:406] Test net output #74: loss2/cross_entropy_loss = 0.866149 (* 0.3 = 0.259845 loss)
I0607 01:02:33.211402 32403 solver.cpp:406] Test net output #75: loss2/cross_entropy_loss_incl_empty = 0.257656 (* 0.3 = 0.0772968 loss)
I0607 01:02:33.211416 32403 solver.cpp:406] Test net output #76: loss2/loss01 = 0.546213 (* 0.0272727 = 0.0148967 loss)
I0607 01:02:33.211429 32403 solver.cpp:406] Test net output #77: loss2/loss02 = 0.770371 (* 0.0272727 = 0.0210101 loss)
I0607 01:02:33.211452 32403 solver.cpp:406] Test net output #78: loss2/loss03 = 0.898445 (* 0.0272727 = 0.024503 loss)
I0607 01:02:33.211472 32403 solver.cpp:406] Test net output #79: loss2/loss04 = 1.15967 (* 0.0272727 = 0.0316275 loss)
I0607 01:02:33.211484 32403 solver.cpp:406] Test net output #80: loss2/loss05 = 1.04251 (* 0.0272727 = 0.028432 loss)
I0607 01:02:33.211498 32403 solver.cpp:406] Test net output #81: loss2/loss06 = 0.722815 (* 0.0272727 = 0.0197131 loss)
I0607 01:02:33.211511 32403 solver.cpp:406] Test net output #82: loss2/loss07 = 0.439547 (* 0.0272727 = 0.0119876 loss)
I0607 01:02:33.211525 32403 solver.cpp:406] Test net output #83: loss2/loss08 = 0.233419 (* 0.0272727 = 0.00636598 loss)
I0607 01:02:33.211539 32403 solver.cpp:406] Test net output #84: loss2/loss09 = 0.140772 (* 0.0272727 = 0.00383923 loss)
I0607 01:02:33.211552 32403 solver.cpp:406] Test net output #85: loss2/loss10 = 0.0697725 (* 0.0272727 = 0.00190289 loss)
I0607 01:02:33.211566 32403 solver.cpp:406] Test net output #86: loss2/loss11 = 0.0209235 (* 0.0272727 = 0.000570641 loss)
I0607 01:02:33.211580 32403 solver.cpp:406] Test net output #87: loss2/loss12 = 0.0118287 (* 0.0272727 = 0.000322602 loss)
I0607 01:02:33.211594 32403 solver.cpp:406] Test net output #88: loss2/loss13 = 0.00817795 (* 0.0272727 = 0.000223035 loss)
I0607 01:02:33.211608 32403 solver.cpp:406] Test net output #89: loss2/loss14 = 0.00587035 (* 0.0272727 = 0.0001601 loss)
I0607 01:02:33.211621 32403 solver.cpp:406] Test net output #90: loss2/loss15 = 0.00402254 (* 0.0272727 = 0.000109706 loss)
I0607 01:02:33.211635 32403 solver.cpp:406] Test net output #91: loss2/loss16 = 0.00306248 (* 0.0272727 = 8.35221e-05 loss)
I0607 01:02:33.211649 32403 solver.cpp:406] Test net output #92: loss2/loss17 = 0.00250988 (* 0.0272727 = 6.84513e-05 loss)
I0607 01:02:33.211663 32403 solver.cpp:406] Test net output #93: loss2/loss18 = 0.00215353 (* 0.0272727 = 5.87328e-05 loss)
I0607 01:02:33.211676 32403 solver.cpp:406] Test net output #94: loss2/loss19 = 0.00162662 (* 0.0272727 = 4.43623e-05 loss)
I0607 01:02:33.211690 32403 solver.cpp:406] Test net output #95: loss2/loss20 = 0.00133587 (* 0.0272727 = 3.64329e-05 loss)
I0607 01:02:33.211704 32403 solver.cpp:406] Test net output #96: loss2/loss21 = 0.000984625 (* 0.0272727 = 2.68534e-05 loss)
I0607 01:02:33.211717 32403 solver.cpp:406] Test net output #97: loss2/loss22 = 0.000710294 (* 0.0272727 = 1.93717e-05 loss)
I0607 01:02:33.211730 32403 solver.cpp:406] Test net output #98: loss3/accuracy = 0.861763
I0607 01:02:33.211741 32403 solver.cpp:406] Test net output #99: loss3/accuracy01 = 0.897
I0607 01:02:33.211752 32403 solver.cpp:406] Test net output #100: loss3/accuracy02 = 0.873
I0607 01:02:33.211765 32403 solver.cpp:406] Test net output #101: loss3/accuracy03 = 0.872
I0607 01:02:33.211776 32403 solver.cpp:406] Test net output #102: loss3/accuracy04 = 0.858
I0607 01:02:33.211786 32403 solver.cpp:406] Test net output #103: loss3/accuracy05 = 0.85
I0607 01:02:33.211798 32403 solver.cpp:406] Test net output #104: loss3/accuracy06 = 0.874
I0607 01:02:33.211809 32403 solver.cpp:406] Test net output #105: loss3/accuracy07 = 0.909
I0607 01:02:33.211820 32403 solver.cpp:406] Test net output #106: loss3/accuracy08 = 0.944
I0607 01:02:33.211832 32403 solver.cpp:406] Test net output #107: loss3/accuracy09 = 0.971
I0607 01:02:33.211843 32403 solver.cpp:406] Test net output #108: loss3/accuracy10 = 0.983
I0607 01:02:33.211854 32403 solver.cpp:406] Test net output #109: loss3/accuracy11 = 0.995
I0607 01:02:33.211865 32403 solver.cpp:406] Test net output #110: loss3/accuracy12 = 0.997
I0607 01:02:33.211876 32403 solver.cpp:406] Test net output #111: loss3/accuracy13 = 0.999
I0607 01:02:33.211889 32403 solver.cpp:406] Test net output #112: loss3/accuracy14 = 0.999
I0607 01:02:33.211902 32403 solver.cpp:406] Test net output #113: loss3/accuracy15 = 1
I0607 01:02:33.211910 32403 solver.cpp:406] Test net output #114: loss3/accuracy16 = 1
I0607 01:02:33.211932 32403 solver.cpp:406] Test net output #115: loss3/accuracy17 = 1
I0607 01:02:33.211946 32403 solver.cpp:406] Test net output #116: loss3/accuracy18 = 1
I0607 01:02:33.211958 32403 solver.cpp:406] Test net output #117: loss3/accuracy19 = 1
I0607 01:02:33.211969 32403 solver.cpp:406] Test net output #118: loss3/accuracy20 = 1
I0607 01:02:33.211980 32403 solver.cpp:406] Test net output #119: loss3/accuracy21 = 1
I0607 01:02:33.211992 32403 solver.cpp:406] Test net output #120: loss3/accuracy22 = 1
I0607 01:02:33.212002 32403 solver.cpp:406] Test net output #121: loss3/accuracy_incl_empty = 0.955546
I0607 01:02:33.212013 32403 solver.cpp:406] Test net output #122: loss3/accuracy_top3 = 0.935263
I0607 01:02:33.212028 32403 solver.cpp:406] Test net output #123: loss3/cross_entropy_loss = 0.644178 (* 1 = 0.644178 loss)
I0607 01:02:33.212040 32403 solver.cpp:406] Test net output #124: loss3/cross_entropy_loss_incl_empty = 0.205174 (* 1 = 0.205174 loss)
I0607 01:02:33.212054 32403 solver.cpp:406] Test net output #125: loss3/loss01 = 0.49672 (* 0.0909091 = 0.0451563 loss)
I0607 01:02:33.212069 32403 solver.cpp:406] Test net output #126: loss3/loss02 = 0.602839 (* 0.0909091 = 0.0548036 loss)
I0607 01:02:33.212081 32403 solver.cpp:406] Test net output #127: loss3/loss03 = 0.615064 (* 0.0909091 = 0.0559149 loss)
I0607 01:02:33.212095 32403 solver.cpp:406] Test net output #128: loss3/loss04 = 0.678051 (* 0.0909091 = 0.061641 loss)
I0607 01:02:33.212108 32403 solver.cpp:406] Test net output #129: loss3/loss05 = 0.66459 (* 0.0909091 = 0.0604172 loss)
I0607 01:02:33.212121 32403 solver.cpp:406] Test net output #130: loss3/loss06 = 0.542878 (* 0.0909091 = 0.0493525 loss)
I0607 01:02:33.212136 32403 solver.cpp:406] Test net output #131: loss3/loss07 = 0.37971 (* 0.0909091 = 0.0345191 loss)
I0607 01:02:33.212148 32403 solver.cpp:406] Test net output #132: loss3/loss08 = 0.207042 (* 0.0909091 = 0.018822 loss)
I0607 01:02:33.212162 32403 solver.cpp:406] Test net output #133: loss3/loss09 = 0.122922 (* 0.0909091 = 0.0111747 loss)
I0607 01:02:33.212175 32403 solver.cpp:406] Test net output #134: loss3/loss10 = 0.072653 (* 0.0909091 = 0.00660482 loss)
I0607 01:02:33.212188 32403 solver.cpp:406] Test net output #135: loss3/loss11 = 0.0203253 (* 0.0909091 = 0.00184775 loss)
I0607 01:02:33.212201 32403 solver.cpp:406] Test net output #136: loss3/loss12 = 0.0110913 (* 0.0909091 = 0.0010083 loss)
I0607 01:02:33.212215 32403 solver.cpp:406] Test net output #137: loss3/loss13 = 0.00709283 (* 0.0909091 = 0.000644803 loss)
I0607 01:02:33.212229 32403 solver.cpp:406] Test net output #138: loss3/loss14 = 0.00444924 (* 0.0909091 = 0.000404477 loss)
I0607 01:02:33.212242 32403 solver.cpp:406] Test net output #139: loss3/loss15 = 0.00231933 (* 0.0909091 = 0.000210848 loss)
I0607 01:02:33.212255 32403 solver.cpp:406] Test net output #140: loss3/loss16 = 0.00122322 (* 0.0909091 = 0.000111202 loss)
I0607 01:02:33.212268 32403 solver.cpp:406] Test net output #141: loss3/loss17 = 0.00103402 (* 0.0909091 = 9.40018e-05 loss)
I0607 01:02:33.212281 32403 solver.cpp:406] Test net output #142: loss3/loss18 = 0.000745559 (* 0.0909091 = 6.77781e-05 loss)
I0607 01:02:33.212296 32403 solver.cpp:406] Test net output #143: loss3/loss19 = 0.000620666 (* 0.0909091 = 5.64242e-05 loss)
I0607 01:02:33.212308 32403 solver.cpp:406] Test net output #144: loss3/loss20 = 0.00048764 (* 0.0909091 = 4.43309e-05 loss)
I0607 01:02:33.212322 32403 solver.cpp:406] Test net output #145: loss3/loss21 = 0.000357319 (* 0.0909091 = 3.24835e-05 loss)
I0607 01:02:33.212335 32403 solver.cpp:406] Test net output #146: loss3/loss22 = 0.000306961 (* 0.0909091 = 2.79056e-05 loss)
I0607 01:02:33.212347 32403 solver.cpp:406] Test net output #147: total_accuracy = 0.624
I0607 01:02:33.212358 32403 solver.cpp:406] Test net output #148: total_accuracy_not_rec = 0.6
I0607 01:02:33.212369 32403 solver.cpp:406] Test net output #149: total_confidence = 0.606343
I0607 01:02:33.212390 32403 solver.cpp:406] Test net output #150: total_confidence_not_rec = 0.550891
I0607 01:02:33.212404 32403 solver.cpp:338] Iteration 0, Testing net (#1)
I0607 01:03:33.542515 32403 solver.cpp:393] Test loss: 3.69487
I0607 01:03:33.542973 32403 solver.cpp:406] Test net output #0: loss1/accuracy = 0.573623
I0607 01:03:33.543000 32403 solver.cpp:406] Test net output #1: loss1/accuracy01 = 0.768
I0607 01:03:33.543014 32403 solver.cpp:406] Test net output #2: loss1/accuracy02 = 0.671
I0607 01:03:33.543025 32403 solver.cpp:406] Test net output #3: loss1/accuracy03 = 0.539
I0607 01:03:33.543037 32403 solver.cpp:406] Test net output #4: loss1/accuracy04 = 0.415
I0607 01:03:33.543053 32403 solver.cpp:406] Test net output #5: loss1/accuracy05 = 0.489
I0607 01:03:33.543066 32403 solver.cpp:406] Test net output #6: loss1/accuracy06 = 0.63
I0607 01:03:33.543077 32403 solver.cpp:406] Test net output #7: loss1/accuracy07 = 0.75
I0607 01:03:33.543089 32403 solver.cpp:406] Test net output #8: loss1/accuracy08 = 0.801
I0607 01:03:33.543102 32403 solver.cpp:406] Test net output #9: loss1/accuracy09 = 0.832
I0607 01:03:33.543117 32403 solver.cpp:406] Test net output #10: loss1/accuracy10 = 0.852
I0607 01:03:33.543128 32403 solver.cpp:406] Test net output #11: loss1/accuracy11 = 0.894
I0607 01:03:33.543140 32403 solver.cpp:406] Test net output #12: loss1/accuracy12 = 0.896
I0607 01:03:33.543153 32403 solver.cpp:406] Test net output #13: loss1/accuracy13 = 0.923
I0607 01:03:33.543164 32403 solver.cpp:406] Test net output #14: loss1/accuracy14 = 0.936
I0607 01:03:33.543176 32403 solver.cpp:406] Test net output #15: loss1/accuracy15 = 0.952
I0607 01:03:33.543192 32403 solver.cpp:406] Test net output #16: loss1/accuracy16 = 0.967
I0607 01:03:33.543205 32403 solver.cpp:406] Test net output #17: loss1/accuracy17 = 0.983
I0607 01:03:33.543215 32403 solver.cpp:406] Test net output #18: loss1/accuracy18 = 0.988
I0607 01:03:33.543227 32403 solver.cpp:406] Test net output #19: loss1/accuracy19 = 0.99
I0607 01:03:33.543238 32403 solver.cpp:406] Test net output #20: loss1/accuracy20 = 0.996
I0607 01:03:33.543254 32403 solver.cpp:406] Test net output #21: loss1/accuracy21 = 0.999
I0607 01:03:33.543267 32403 solver.cpp:406] Test net output #22: loss1/accuracy22 = 1
I0607 01:03:33.543277 32403 solver.cpp:406] Test net output #23: loss1/accuracy_incl_empty = 0.840774
I0607 01:03:33.543289 32403 solver.cpp:406] Test net output #24: loss1/accuracy_top3 = 0.795894
I0607 01:03:33.543305 32403 solver.cpp:406] Test net output #25: loss1/cross_entropy_loss = 1.63711 (* 0.3 = 0.491132 loss)
I0607 01:03:33.543320 32403 solver.cpp:406] Test net output #26: loss1/cross_entropy_loss_incl_empty = 0.602438 (* 0.3 = 0.180731 loss)
I0607 01:03:33.543334 32403 solver.cpp:406] Test net output #27: loss1/loss01 = 1.01529 (* 0.0272727 = 0.0276899 loss)
I0607 01:03:33.543349 32403 solver.cpp:406] Test net output #28: loss1/loss02 = 1.36126 (* 0.0272727 = 0.0371253 loss)
I0607 01:03:33.543361 32403 solver.cpp:406] Test net output #29: loss1/loss03 = 1.72643 (* 0.0272727 = 0.0470843 loss)
I0607 01:03:33.543375 32403 solver.cpp:406] Test net output #30: loss1/loss04 = 1.99734 (* 0.0272727 = 0.0544728 loss)
I0607 01:03:33.543388 32403 solver.cpp:406] Test net output #31: loss1/loss05 = 1.73764 (* 0.0272727 = 0.0473903 loss)
I0607 01:03:33.543402 32403 solver.cpp:406] Test net output #32: loss1/loss06 = 1.2956 (* 0.0272727 = 0.0353345 loss)
I0607 01:03:33.543416 32403 solver.cpp:406] Test net output #33: loss1/loss07 = 0.89027 (* 0.0272727 = 0.0242801 loss)
I0607 01:03:33.543429 32403 solver.cpp:406] Test net output #34: loss1/loss08 = 0.724801 (* 0.0272727 = 0.0197673 loss)
I0607 01:03:33.543443 32403 solver.cpp:406] Test net output #35: loss1/loss09 = 0.602161 (* 0.0272727 = 0.0164226 loss)
I0607 01:03:33.543457 32403 solver.cpp:406] Test net output #36: loss1/loss10 = 0.531203 (* 0.0272727 = 0.0144874 loss)
I0607 01:03:33.543470 32403 solver.cpp:406] Test net output #37: loss1/loss11 = 0.458899 (* 0.0272727 = 0.0125154 loss)
I0607 01:03:33.543483 32403 solver.cpp:406] Test net output #38: loss1/loss12 = 0.383589 (* 0.0272727 = 0.0104615 loss)
I0607 01:03:33.543519 32403 solver.cpp:406] Test net output #39: loss1/loss13 = 0.275785 (* 0.0272727 = 0.0075214 loss)
I0607 01:03:33.543535 32403 solver.cpp:406] Test net output #40: loss1/loss14 = 0.234865 (* 0.0272727 = 0.00640541 loss)
I0607 01:03:33.543548 32403 solver.cpp:406] Test net output #41: loss1/loss15 = 0.182166 (* 0.0272727 = 0.00496817 loss)
I0607 01:03:33.543562 32403 solver.cpp:406] Test net output #42: loss1/loss16 = 0.146109 (* 0.0272727 = 0.00398478 loss)
I0607 01:03:33.543576 32403 solver.cpp:406] Test net output #43: loss1/loss17 = 0.0926692 (* 0.0272727 = 0.00252734 loss)
I0607 01:03:33.543591 32403 solver.cpp:406] Test net output #44: loss1/loss18 = 0.0690393 (* 0.0272727 = 0.00188289 loss)
I0607 01:03:33.543606 32403 solver.cpp:406] Test net output #45: loss1/loss19 = 0.0585716 (* 0.0272727 = 0.00159741 loss)
I0607 01:03:33.543624 32403 solver.cpp:406] Test net output #46: loss1/loss20 = 0.0254355 (* 0.0272727 = 0.000693695 loss)
I0607 01:03:33.543638 32403 solver.cpp:406] Test net output #47: loss1/loss21 = 0.00872184 (* 0.0272727 = 0.000237868 loss)
I0607 01:03:33.543653 32403 solver.cpp:406] Test net output #48: loss1/loss22 = 0.00153605 (* 0.0272727 = 4.18923e-05 loss)
I0607 01:03:33.543664 32403 solver.cpp:406] Test net output #49: loss2/accuracy = 0.710684
I0607 01:03:33.543678 32403 solver.cpp:406] Test net output #50: loss2/accuracy01 = 0.86
I0607 01:03:33.543689 32403 solver.cpp:406] Test net output #51: loss2/accuracy02 = 0.82
I0607 01:03:33.543699 32403 solver.cpp:406] Test net output #52: loss2/accuracy03 = 0.75
I0607 01:03:33.543711 32403 solver.cpp:406] Test net output #53: loss2/accuracy04 = 0.637
I0607 01:03:33.543722 32403 solver.cpp:406] Test net output #54: loss2/accuracy05 = 0.623
I0607 01:03:33.543735 32403 solver.cpp:406] Test net output #55: loss2/accuracy06 = 0.729
I0607 01:03:33.543745 32403 solver.cpp:406] Test net output #56: loss2/accuracy07 = 0.786
I0607 01:03:33.543756 32403 solver.cpp:406] Test net output #57: loss2/accuracy08 = 0.832
I0607 01:03:33.543768 32403 solver.cpp:406] Test net output #58: loss2/accuracy09 = 0.848
I0607 01:03:33.543779 32403 solver.cpp:406] Test net output #59: loss2/accuracy10 = 0.87
I0607 01:03:33.543792 32403 solver.cpp:406] Test net output #60: loss2/accuracy11 = 0.895
I0607 01:03:33.543802 32403 solver.cpp:406] Test net output #61: loss2/accuracy12 = 0.904
I0607 01:03:33.543813 32403 solver.cpp:406] Test net output #62: loss2/accuracy13 = 0.936
I0607 01:03:33.543825 32403 solver.cpp:406] Test net output #63: loss2/accuracy14 = 0.939
I0607 01:03:33.543836 32403 solver.cpp:406] Test net output #64: loss2/accuracy15 = 0.953
I0607 01:03:33.543848 32403 solver.cpp:406] Test net output #65: loss2/accuracy16 = 0.969
I0607 01:03:33.543859 32403 solver.cpp:406] Test net output #66: loss2/accuracy17 = 0.983
I0607 01:03:33.543870 32403 solver.cpp:406] Test net output #67: loss2/accuracy18 = 0.988
I0607 01:03:33.543882 32403 solver.cpp:406] Test net output #68: loss2/accuracy19 = 0.99
I0607 01:03:33.543894 32403 solver.cpp:406] Test net output #69: loss2/accuracy20 = 0.996
I0607 01:03:33.543905 32403 solver.cpp:406] Test net output #70: loss2/accuracy21 = 0.999
I0607 01:03:33.543916 32403 solver.cpp:406] Test net output #71: loss2/accuracy22 = 1
I0607 01:03:33.543931 32403 solver.cpp:406] Test net output #72: loss2/accuracy_incl_empty = 0.883047
I0607 01:03:33.543942 32403 solver.cpp:406] Test net output #73: loss2/accuracy_top3 = 0.868504
I0607 01:03:33.543956 32403 solver.cpp:406] Test net output #74: loss2/cross_entropy_loss = 1.201 (* 0.3 = 0.360299 loss)
I0607 01:03:33.543969 32403 solver.cpp:406] Test net output #75: loss2/cross_entropy_loss_incl_empty = 0.459492 (* 0.3 = 0.137847 loss)
I0607 01:03:33.543983 32403 solver.cpp:406] Test net output #76: loss2/loss01 = 0.734171 (* 0.0272727 = 0.0200228 loss)
I0607 01:03:33.543998 32403 solver.cpp:406] Test net output #77: loss2/loss02 = 0.893073 (* 0.0272727 = 0.0243565 loss)
I0607 01:03:33.544021 32403 solver.cpp:406] Test net output #78: loss2/loss03 = 1.05792 (* 0.0272727 = 0.0288524 loss)
I0607 01:03:33.544036 32403 solver.cpp:406] Test net output #79: loss2/loss04 = 1.32342 (* 0.0272727 = 0.0360933 loss)
I0607 01:03:33.544050 32403 solver.cpp:406] Test net output #80: loss2/loss05 = 1.29214 (* 0.0272727 = 0.0352403 loss)
I0607 01:03:33.544064 32403 solver.cpp:406] Test net output #81: loss2/loss06 = 0.995269 (* 0.0272727 = 0.0271437 loss)
I0607 01:03:33.544076 32403 solver.cpp:406] Test net output #82: loss2/loss07 = 0.778856 (* 0.0272727 = 0.0212415 loss)
I0607 01:03:33.544090 32403 solver.cpp:406] Test net output #83: loss2/loss08 = 0.638771 (* 0.0272727 = 0.017421 loss)
I0607 01:03:33.544104 32403 solver.cpp:406] Test net output #84: loss2/loss09 = 0.546223 (* 0.0272727 = 0.014897 loss)
I0607 01:03:33.544117 32403 solver.cpp:406] Test net output #85: loss2/loss10 = 0.476832 (* 0.0272727 = 0.0130045 loss)
I0607 01:03:33.544131 32403 solver.cpp:406] Test net output #86: loss2/loss11 = 0.414523 (* 0.0272727 = 0.0113052 loss)
I0607 01:03:33.544147 32403 solver.cpp:406] Test net output #87: loss2/loss12 = 0.335205 (* 0.0272727 = 0.00914195 loss)
I0607 01:03:33.544160 32403 solver.cpp:406] Test net output #88: loss2/loss13 = 0.244797 (* 0.0272727 = 0.00667628 loss)
I0607 01:03:33.544178 32403 solver.cpp:406] Test net output #89: loss2/loss14 = 0.212235 (* 0.0272727 = 0.00578822 loss)
I0607 01:03:33.544193 32403 solver.cpp:406] Test net output #90: loss2/loss15 = 0.174367 (* 0.0272727 = 0.00475546 loss)
I0607 01:03:33.544206 32403 solver.cpp:406] Test net output #91: loss2/loss16 = 0.141715 (* 0.0272727 = 0.00386496 loss)
I0607 01:03:33.544220 32403 solver.cpp:406] Test net output #92: loss2/loss17 = 0.0891539 (* 0.0272727 = 0.00243147 loss)
I0607 01:03:33.544234 32403 solver.cpp:406] Test net output #93: loss2/loss18 = 0.0624149 (* 0.0272727 = 0.00170223 loss)
I0607 01:03:33.544247 32403 solver.cpp:406] Test net output #94: loss2/loss19 = 0.0604684 (* 0.0272727 = 0.00164914 loss)
I0607 01:03:33.544261 32403 solver.cpp:406] Test net output #95: loss2/loss20 = 0.0257822 (* 0.0272727 = 0.00070315 loss)
I0607 01:03:33.544275 32403 solver.cpp:406] Test net output #96: loss2/loss21 = 0.00874604 (* 0.0272727 = 0.000238528 loss)
I0607 01:03:33.544288 32403 solver.cpp:406] Test net output #97: loss2/loss22 = 0.000791085 (* 0.0272727 = 2.1575e-05 loss)
I0607 01:03:33.544301 32403 solver.cpp:406] Test net output #98: loss3/accuracy = 0.814677
I0607 01:03:33.544312 32403 solver.cpp:406] Test net output #99: loss3/accuracy01 = 0.874
I0607 01:03:33.544323 32403 solver.cpp:406] Test net output #100: loss3/accuracy02 = 0.852
I0607 01:03:33.544332 32403 solver.cpp:406] Test net output #101: loss3/accuracy03 = 0.848
I0607 01:03:33.544339 32403 solver.cpp:406] Test net output #102: loss3/accuracy04 = 0.835
I0607 01:03:33.544350 32403 solver.cpp:406] Test net output #103: loss3/accuracy05 = 0.826
I0607 01:03:33.544363 32403 solver.cpp:406] Test net output #104: loss3/accuracy06 = 0.841
I0607 01:03:33.544374 32403 solver.cpp:406] Test net output #105: loss3/accuracy07 = 0.88
I0607 01:03:33.544385 32403 solver.cpp:406] Test net output #106: loss3/accuracy08 = 0.89
I0607 01:03:33.544397 32403 solver.cpp:406] Test net output #107: loss3/accuracy09 = 0.907
I0607 01:03:33.544409 32403 solver.cpp:406] Test net output #108: loss3/accuracy10 = 0.91
I0607 01:03:33.544420 32403 solver.cpp:406] Test net output #109: loss3/accuracy11 = 0.924
I0607 01:03:33.544431 32403 solver.cpp:406] Test net output #110: loss3/accuracy12 = 0.926
I0607 01:03:33.544442 32403 solver.cpp:406] Test net output #111: loss3/accuracy13 = 0.942
I0607 01:03:33.544453 32403 solver.cpp:406] Test net output #112: loss3/accuracy14 = 0.961
I0607 01:03:33.544466 32403 solver.cpp:406] Test net output #113: loss3/accuracy15 = 0.964
I0607 01:03:33.544476 32403 solver.cpp:406] Test net output #114: loss3/accuracy16 = 0.974
I0607 01:03:33.544498 32403 solver.cpp:406] Test net output #115: loss3/accuracy17 = 0.984
I0607 01:03:33.544512 32403 solver.cpp:406] Test net output #116: loss3/accuracy18 = 0.991
I0607 01:03:33.544523 32403 solver.cpp:406] Test net output #117: loss3/accuracy19 = 0.992
I0607 01:03:33.544534 32403 solver.cpp:406] Test net output #118: loss3/accuracy20 = 0.996
I0607 01:03:33.544545 32403 solver.cpp:406] Test net output #119: loss3/accuracy21 = 0.999
I0607 01:03:33.544558 32403 solver.cpp:406] Test net output #120: loss3/accuracy22 = 1
I0607 01:03:33.544569 32403 solver.cpp:406] Test net output #121: loss3/accuracy_incl_empty = 0.922183
I0607 01:03:33.544580 32403 solver.cpp:406] Test net output #122: loss3/accuracy_top3 = 0.908757
I0607 01:03:33.544594 32403 solver.cpp:406] Test net output #123: loss3/cross_entropy_loss = 0.854211 (* 1 = 0.854211 loss)
I0607 01:03:33.544607 32403 solver.cpp:406] Test net output #124: loss3/cross_entropy_loss_incl_empty = 0.332819 (* 1 = 0.332819 loss)
I0607 01:03:33.544620 32403 solver.cpp:406] Test net output #125: loss3/loss01 = 0.63742 (* 0.0909091 = 0.0579473 loss)
I0607 01:03:33.544634 32403 solver.cpp:406] Test net output #126: loss3/loss02 = 0.740147 (* 0.0909091 = 0.0672861 loss)
I0607 01:03:33.544647 32403 solver.cpp:406] Test net output #127: loss3/loss03 = 0.748333 (* 0.0909091 = 0.0680303 loss)
I0607 01:03:33.544661 32403 solver.cpp:406] Test net output #128: loss3/loss04 = 0.761518 (* 0.0909091 = 0.0692289 loss)
I0607 01:03:33.544674 32403 solver.cpp:406] Test net output #129: loss3/loss05 = 0.790507 (* 0.0909091 = 0.0718643 loss)
I0607 01:03:33.544687 32403 solver.cpp:406] Test net output #130: loss3/loss06 = 0.654509 (* 0.0909091 = 0.0595008 loss)
I0607 01:03:33.544701 32403 solver.cpp:406] Test net output #131: loss3/loss07 = 0.525013 (* 0.0909091 = 0.0477285 loss)
I0607 01:03:33.544714 32403 solver.cpp:406] Test net output #132: loss3/loss08 = 0.481048 (* 0.0909091 = 0.0437317 loss)
I0607 01:03:33.544728 32403 solver.cpp:406] Test net output #133: loss3/loss09 = 0.384789 (* 0.0909091 = 0.0349808 loss)
I0607 01:03:33.544741 32403 solver.cpp:406] Test net output #134: loss3/loss10 = 0.347132 (* 0.0909091 = 0.0315574 loss)
I0607 01:03:33.544754 32403 solver.cpp:406] Test net output #135: loss3/loss11 = 0.293162 (* 0.0909091 = 0.0266511 loss)
I0607 01:03:33.544767 32403 solver.cpp:406] Test net output #136: loss3/loss12 = 0.265846 (* 0.0909091 = 0.0241678 loss)
I0607 01:03:33.544781 32403 solver.cpp:406] Test net output #137: loss3/loss13 = 0.208755 (* 0.0909091 = 0.0189778 loss)
I0607 01:03:33.544795 32403 solver.cpp:406] Test net output #138: loss3/loss14 = 0.157002 (* 0.0909091 = 0.0142729 loss)
I0607 01:03:33.544807 32403 solver.cpp:406] Test net output #139: loss3/loss15 = 0.137587 (* 0.0909091 = 0.0125079 loss)
I0607 01:03:33.544821 32403 solver.cpp:406] Test net output #140: loss3/loss16 = 0.106111 (* 0.0909091 = 0.00964641 loss)
I0607 01:03:33.544836 32403 solver.cpp:406] Test net output #141: loss3/loss17 = 0.0707271 (* 0.0909091 = 0.00642974 loss)
I0607 01:03:33.544849 32403 solver.cpp:406] Test net output #142: loss3/loss18 = 0.0447581 (* 0.0909091 = 0.00406892 loss)
I0607 01:03:33.544862 32403 solver.cpp:406] Test net output #143: loss3/loss19 = 0.0405742 (* 0.0909091 = 0.00368857 loss)
I0607 01:03:33.544877 32403 solver.cpp:406] Test net output #144: loss3/loss20 = 0.0169184 (* 0.0909091 = 0.00153803 loss)
I0607 01:03:33.544890 32403 solver.cpp:406] Test net output #145: loss3/loss21 = 0.00538688 (* 0.0909091 = 0.000489717 loss)
I0607 01:03:33.544903 32403 solver.cpp:406] Test net output #146: loss3/loss22 = 0.000969243 (* 0.0909091 = 8.8113e-05 loss)
I0607 01:03:33.544915 32403 solver.cpp:406] Test net output #147: total_accuracy = 0.534
I0607 01:03:33.544926 32403 solver.cpp:406] Test net output #148: total_accuracy_not_rec = 0.498
I0607 01:03:33.544939 32403 solver.cpp:406] Test net output #149: total_confidence = 0.476343
I0607 01:03:33.544958 32403 solver.cpp:406] Test net output #150: total_confidence_not_rec = 0.427408
I0607 01:03:34.009052 32403 solver.cpp:229] Iteration 0, loss = 1.29039
I0607 01:03:34.009132 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.697674
I0607 01:03:34.009153 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.875
I0607 01:03:34.009167 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.625
I0607 01:03:34.009181 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.625
I0607 01:03:34.009193 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.5
I0607 01:03:34.009205 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.625
I0607 01:03:34.009217 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.875
I0607 01:03:34.009232 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.875
I0607 01:03:34.009244 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 1
I0607 01:03:34.009256 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 1
I0607 01:03:34.009268 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 1
I0607 01:03:34.009280 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 1
I0607 01:03:34.009292 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 1
I0607 01:03:34.009305 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 1
I0607 01:03:34.009316 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 1
I0607 01:03:34.009328 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 01:03:34.009340 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 01:03:34.009351 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 01:03:34.009363 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 01:03:34.009376 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 01:03:34.009387 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 01:03:34.009398 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 01:03:34.009409 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 01:03:34.009421 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.920455
I0607 01:03:34.009434 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.837209
I0607 01:03:34.009451 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.0773 (* 0.3 = 0.323191 loss)
I0607 01:03:34.009466 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.281175 (* 0.3 = 0.0843526 loss)
I0607 01:03:34.009481 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.401794 (* 0.0272727 = 0.010958 loss)
I0607 01:03:34.009495 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 0.912902 (* 0.0272727 = 0.0248973 loss)
I0607 01:03:34.009511 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.37612 (* 0.0272727 = 0.0375307 loss)
I0607 01:03:34.009526 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.62447 (* 0.0272727 = 0.0443037 loss)
I0607 01:03:34.009539 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.30585 (* 0.0272727 = 0.0356141 loss)
I0607 01:03:34.009553 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 0.547403 (* 0.0272727 = 0.0149292 loss)
I0607 01:03:34.009568 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 0.761061 (* 0.0272727 = 0.0207562 loss)
I0607 01:03:34.009583 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.0231549 (* 0.0272727 = 0.000631498 loss)
I0607 01:03:34.009598 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.00259948 (* 0.0272727 = 7.08948e-05 loss)
I0607 01:03:34.009613 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.000510955 (* 0.0272727 = 1.39351e-05 loss)
I0607 01:03:34.009626 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.000380208 (* 0.0272727 = 1.03693e-05 loss)
I0607 01:03:34.009683 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.000367879 (* 0.0272727 = 1.00331e-05 loss)
I0607 01:03:34.009699 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.000193761 (* 0.0272727 = 5.28439e-06 loss)
I0607 01:03:34.009713 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.000129959 (* 0.0272727 = 3.54433e-06 loss)
I0607 01:03:34.009727 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.000114588 (* 0.0272727 = 3.12512e-06 loss)
I0607 01:03:34.009752 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 6.20623e-05 (* 0.0272727 = 1.69261e-06 loss)
I0607 01:03:34.009765 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 6.55512e-05 (* 0.0272727 = 1.78776e-06 loss)
I0607 01:03:34.009779 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 3.13925e-05 (* 0.0272727 = 8.56159e-07 loss)
I0607 01:03:34.009802 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 9.65854e-05 (* 0.0272727 = 2.63415e-06 loss)
I0607 01:03:34.009816 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.000116327 (* 0.0272727 = 3.17255e-06 loss)
I0607 01:03:34.009830 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 5.59727e-05 (* 0.0272727 = 1.52653e-06 loss)
I0607 01:03:34.009845 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 8.66194e-05 (* 0.0272727 = 2.36235e-06 loss)
I0607 01:03:34.009857 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.906977
I0607 01:03:34.009871 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.875
I0607 01:03:34.009882 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 1
I0607 01:03:34.009894 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.875
I0607 01:03:34.009905 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.75
I0607 01:03:34.009918 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.625
I0607 01:03:34.009930 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 1
I0607 01:03:34.009941 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.875
I0607 01:03:34.009953 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 1
I0607 01:03:34.009965 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 1
I0607 01:03:34.009977 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 1
I0607 01:03:34.009989 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 1
I0607 01:03:34.010000 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 1
I0607 01:03:34.010011 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 1
I0607 01:03:34.010023 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 1
I0607 01:03:34.010035 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 01:03:34.010046 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 01:03:34.010063 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 01:03:34.010087 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 01:03:34.010121 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 01:03:34.010135 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 01:03:34.010148 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 01:03:34.010167 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 01:03:34.010179 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.965909
I0607 01:03:34.010191 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.953488
I0607 01:03:34.010206 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 0.450811 (* 0.3 = 0.135243 loss)
I0607 01:03:34.010221 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.132138 (* 0.3 = 0.0396414 loss)
I0607 01:03:34.010248 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.392266 (* 0.0272727 = 0.0106982 loss)
I0607 01:03:34.010264 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.266354 (* 0.0272727 = 0.00726419 loss)
I0607 01:03:34.010282 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.520141 (* 0.0272727 = 0.0141857 loss)
I0607 01:03:34.010296 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 0.820496 (* 0.0272727 = 0.0223772 loss)
I0607 01:03:34.010311 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 0.953922 (* 0.0272727 = 0.0260161 loss)
I0607 01:03:34.010325 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 0.31846 (* 0.0272727 = 0.00868527 loss)
I0607 01:03:34.010339 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.734315 (* 0.0272727 = 0.0200268 loss)
I0607 01:03:34.010354 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.0112005 (* 0.0272727 = 0.000305469 loss)
I0607 01:03:34.010368 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.00273283 (* 0.0272727 = 7.45318e-05 loss)
I0607 01:03:34.010382 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.000276129 (* 0.0272727 = 7.53079e-06 loss)
I0607 01:03:34.010397 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.00103106 (* 0.0272727 = 2.81197e-05 loss)
I0607 01:03:34.010411 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.000224489 (* 0.0272727 = 6.12242e-06 loss)
I0607 01:03:34.010426 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.000127901 (* 0.0272727 = 3.4882e-06 loss)
I0607 01:03:34.010440 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.000196536 (* 0.0272727 = 5.36007e-06 loss)
I0607 01:03:34.010454 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.000155066 (* 0.0272727 = 4.22906e-06 loss)
I0607 01:03:34.010468 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.000201332 (* 0.0272727 = 5.49088e-06 loss)
I0607 01:03:34.010483 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.00039342 (* 0.0272727 = 1.07296e-05 loss)
I0607 01:03:34.010499 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.000205662 (* 0.0272727 = 5.60895e-06 loss)
I0607 01:03:34.010509 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 8.39868e-05 (* 0.0272727 = 2.29055e-06 loss)
I0607 01:03:34.010524 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.000152246 (* 0.0272727 = 4.15216e-06 loss)
I0607 01:03:34.010538 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.000267878 (* 0.0272727 = 7.30576e-06 loss)
I0607 01:03:34.010552 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 7.93282e-05 (* 0.0272727 = 2.1635e-06 loss)
I0607 01:03:34.010565 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.906977
I0607 01:03:34.010577 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 0.875
I0607 01:03:34.010589 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 1
I0607 01:03:34.010601 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.875
I0607 01:03:34.010612 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.875
I0607 01:03:34.010624 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 1
I0607 01:03:34.010637 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.875
I0607 01:03:34.010648 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.875
I0607 01:03:34.010659 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 1
I0607 01:03:34.010671 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 1
I0607 01:03:34.010684 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 1
I0607 01:03:34.010694 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 1
I0607 01:03:34.010706 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 1
I0607 01:03:34.010717 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 1
I0607 01:03:34.010740 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 01:03:34.010753 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 01:03:34.010764 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 01:03:34.010776 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 01:03:34.010788 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 01:03:34.010799 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 01:03:34.010812 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 01:03:34.010823 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 01:03:34.010834 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 01:03:34.010846 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.977273
I0607 01:03:34.010859 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 1
I0607 01:03:34.010872 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.18495 (* 1 = 0.18495 loss)
I0607 01:03:34.010886 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.0486625 (* 1 = 0.0486625 loss)
I0607 01:03:34.010901 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.132336 (* 0.0909091 = 0.0120305 loss)
I0607 01:03:34.010916 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.047355 (* 0.0909091 = 0.004305 loss)
I0607 01:03:34.010931 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.358762 (* 0.0909091 = 0.0326147 loss)
I0607 01:03:34.010944 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.290512 (* 0.0909091 = 0.0264102 loss)
I0607 01:03:34.010959 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.169083 (* 0.0909091 = 0.0153712 loss)
I0607 01:03:34.010973 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.256238 (* 0.0909091 = 0.0232943 loss)
I0607 01:03:34.010987 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.667879 (* 0.0909091 = 0.0607163 loss)
I0607 01:03:34.011001 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.000774794 (* 0.0909091 = 7.04358e-05 loss)
I0607 01:03:34.011016 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 8.57468e-05 (* 0.0909091 = 7.79516e-06 loss)
I0607 01:03:34.011030 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 4.46911e-05 (* 0.0909091 = 4.06283e-06 loss)
I0607 01:03:34.011045 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.000110563 (* 0.0909091 = 1.00512e-05 loss)
I0607 01:03:34.011059 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 6.17293e-05 (* 0.0909091 = 5.61175e-06 loss)
I0607 01:03:34.011073 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.000133382 (* 0.0909091 = 1.21256e-05 loss)
I0607 01:03:34.011088 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 6.65177e-05 (* 0.0909091 = 6.04706e-06 loss)
I0607 01:03:34.011102 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 3.00213e-05 (* 0.0909091 = 2.72921e-06 loss)
I0607 01:03:34.011117 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 3.82486e-05 (* 0.0909091 = 3.47715e-06 loss)
I0607 01:03:34.011132 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 2.4976e-05 (* 0.0909091 = 2.27054e-06 loss)
I0607 01:03:34.011145 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 1.00137e-05 (* 0.0909091 = 9.10337e-07 loss)
I0607 01:03:34.011163 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 1.86418e-05 (* 0.0909091 = 1.69471e-06 loss)
I0607 01:03:34.011178 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 3.81127e-05 (* 0.0909091 = 3.46479e-06 loss)
I0607 01:03:34.011195 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 1.23384e-05 (* 0.0909091 = 1.12167e-06 loss)
I0607 01:03:34.011220 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 6.16913e-06 (* 0.0909091 = 5.6083e-07 loss)
I0607 01:03:34.011242 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.875
I0607 01:03:34.011255 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.875
I0607 01:03:34.011266 32403 solver.cpp:245] Train net output #149: total_confidence = 0.747454
I0607 01:03:34.011278 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.761629
I0607 01:03:34.011302 32403 sgd_solver.cpp:106] Iteration 0, lr = 0.001
I0607 01:04:25.739796 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.4887 > 30) by scale factor 0.895824
I0607 01:05:27.858101 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 75.5002 > 30) by scale factor 0.39735
I0607 01:05:31.821573 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.0572 > 30) by scale factor 0.998098
I0607 01:05:59.324184 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 36.1313 > 30) by scale factor 0.830306
I0607 01:06:41.757544 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.88 > 30) by scale factor 0.860091
I0607 01:07:39.999090 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.2106 > 30) by scale factor 0.931371
I0607 01:08:55.372189 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.8945 > 30) by scale factor 0.9406
I0607 01:09:04.786562 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.0254 > 30) by scale factor 0.908393
I0607 01:10:07.194320 32403 solver.cpp:229] Iteration 500, loss = 4.39301
I0607 01:10:07.194444 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.260274
I0607 01:10:07.194464 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.5
I0607 01:10:07.194478 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.5
I0607 01:10:07.194490 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.25
I0607 01:10:07.194502 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.25
I0607 01:10:07.194514 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.375
I0607 01:10:07.194526 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0
I0607 01:10:07.194538 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.625
I0607 01:10:07.194551 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.625
I0607 01:10:07.194562 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.75
I0607 01:10:07.194574 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.75
I0607 01:10:07.194586 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.625
I0607 01:10:07.194598 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 01:10:07.194610 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 1
I0607 01:10:07.194622 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.875
I0607 01:10:07.194634 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 0.875
I0607 01:10:07.194655 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 0.875
I0607 01:10:07.194667 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 0.875
I0607 01:10:07.194679 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 0.875
I0607 01:10:07.194695 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 01:10:07.194715 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 01:10:07.194735 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 01:10:07.194752 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 01:10:07.194764 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.670455
I0607 01:10:07.194777 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.547945
I0607 01:10:07.194800 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 2.51263 (* 0.3 = 0.753788 loss)
I0607 01:10:07.194815 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 1.16717 (* 0.3 = 0.350152 loss)
I0607 01:10:07.194829 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 1.51405 (* 0.0272727 = 0.0412923 loss)
I0607 01:10:07.194844 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.83248 (* 0.0272727 = 0.0499768 loss)
I0607 01:10:07.194857 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.97236 (* 0.0272727 = 0.0537916 loss)
I0607 01:10:07.194871 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 2.62366 (* 0.0272727 = 0.0715544 loss)
I0607 01:10:07.194885 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 2.99918 (* 0.0272727 = 0.0817959 loss)
I0607 01:10:07.194900 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 2.98657 (* 0.0272727 = 0.081452 loss)
I0607 01:10:07.194913 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 1.53179 (* 0.0272727 = 0.0417761 loss)
I0607 01:10:07.194927 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 1.09219 (* 0.0272727 = 0.0297869 loss)
I0607 01:10:07.194941 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 1.1241 (* 0.0272727 = 0.0306574 loss)
I0607 01:10:07.194958 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.891672 (* 0.0272727 = 0.0243183 loss)
I0607 01:10:07.194973 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 1.42039 (* 0.0272727 = 0.0387379 loss)
I0607 01:10:07.194988 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.81599 (* 0.0272727 = 0.0222543 loss)
I0607 01:10:07.195021 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.561933 (* 0.0272727 = 0.0153255 loss)
I0607 01:10:07.195037 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.737795 (* 0.0272727 = 0.0201217 loss)
I0607 01:10:07.195051 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.667455 (* 0.0272727 = 0.0182033 loss)
I0607 01:10:07.195065 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.448395 (* 0.0272727 = 0.012229 loss)
I0607 01:10:07.195080 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.596715 (* 0.0272727 = 0.016274 loss)
I0607 01:10:07.195094 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.89768 (* 0.0272727 = 0.0244822 loss)
I0607 01:10:07.195108 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.0118587 (* 0.0272727 = 0.000323419 loss)
I0607 01:10:07.195122 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.006301 (* 0.0272727 = 0.000171845 loss)
I0607 01:10:07.195137 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.00544702 (* 0.0272727 = 0.000148555 loss)
I0607 01:10:07.195152 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.00255639 (* 0.0272727 = 6.97198e-05 loss)
I0607 01:10:07.195163 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.424658
I0607 01:10:07.195175 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.625
I0607 01:10:07.195188 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.75
I0607 01:10:07.195200 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.375
I0607 01:10:07.195211 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.25
I0607 01:10:07.195224 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0
I0607 01:10:07.195235 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.125
I0607 01:10:07.195246 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.75
I0607 01:10:07.195258 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.625
I0607 01:10:07.195271 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.75
I0607 01:10:07.195281 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.625
I0607 01:10:07.195293 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.625
I0607 01:10:07.195304 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.75
I0607 01:10:07.195317 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 1
I0607 01:10:07.195327 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.875
I0607 01:10:07.195339 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 0.875
I0607 01:10:07.195351 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 0.875
I0607 01:10:07.195363 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 0.875
I0607 01:10:07.195375 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 0.875
I0607 01:10:07.195386 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 01:10:07.195399 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 01:10:07.195410 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 01:10:07.195421 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 01:10:07.195433 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.732955
I0607 01:10:07.195446 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.616438
I0607 01:10:07.195459 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 2.06068 (* 0.3 = 0.618204 loss)
I0607 01:10:07.195472 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.957475 (* 0.3 = 0.287243 loss)
I0607 01:10:07.195487 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 1.42124 (* 0.0272727 = 0.0387612 loss)
I0607 01:10:07.195500 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.902376 (* 0.0272727 = 0.0246102 loss)
I0607 01:10:07.195525 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 1.87784 (* 0.0272727 = 0.0512137 loss)
I0607 01:10:07.195540 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 2.44075 (* 0.0272727 = 0.0665659 loss)
I0607 01:10:07.195554 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 3.32729 (* 0.0272727 = 0.0907442 loss)
I0607 01:10:07.195569 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 2.12379 (* 0.0272727 = 0.0579216 loss)
I0607 01:10:07.195581 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 1.27492 (* 0.0272727 = 0.0347705 loss)
I0607 01:10:07.195595 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.868266 (* 0.0272727 = 0.02368 loss)
I0607 01:10:07.195616 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.970659 (* 0.0272727 = 0.0264725 loss)
I0607 01:10:07.195631 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 1.0391 (* 0.0272727 = 0.0283392 loss)
I0607 01:10:07.195644 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 1.32988 (* 0.0272727 = 0.0362693 loss)
I0607 01:10:07.195658 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.646359 (* 0.0272727 = 0.017628 loss)
I0607 01:10:07.195672 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.433938 (* 0.0272727 = 0.0118347 loss)
I0607 01:10:07.195685 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.495764 (* 0.0272727 = 0.0135208 loss)
I0607 01:10:07.195708 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.537369 (* 0.0272727 = 0.0146555 loss)
I0607 01:10:07.195719 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.604847 (* 0.0272727 = 0.0164958 loss)
I0607 01:10:07.195727 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.531862 (* 0.0272727 = 0.0145053 loss)
I0607 01:10:07.195746 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.769863 (* 0.0272727 = 0.0209963 loss)
I0607 01:10:07.195760 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.0166707 (* 0.0272727 = 0.000454654 loss)
I0607 01:10:07.195775 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.0174779 (* 0.0272727 = 0.00047667 loss)
I0607 01:10:07.195788 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.0227097 (* 0.0272727 = 0.000619356 loss)
I0607 01:10:07.195808 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 0.021806 (* 0.0272727 = 0.000594709 loss)
I0607 01:10:07.195822 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.506849
I0607 01:10:07.195833 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 0.75
I0607 01:10:07.195844 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.75
I0607 01:10:07.195857 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.5
I0607 01:10:07.195868 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.5
I0607 01:10:07.195879 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.25
I0607 01:10:07.195891 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.625
I0607 01:10:07.195902 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.75
I0607 01:10:07.195914 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.625
I0607 01:10:07.195926 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.875
I0607 01:10:07.195937 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.75
I0607 01:10:07.195948 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.625
I0607 01:10:07.195961 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.75
I0607 01:10:07.195973 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 1
I0607 01:10:07.195984 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 0.875
I0607 01:10:07.195999 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 0.875
I0607 01:10:07.196012 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 0.875
I0607 01:10:07.196033 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 0.875
I0607 01:10:07.196046 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 0.875
I0607 01:10:07.196058 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 01:10:07.196070 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 01:10:07.196081 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 01:10:07.196094 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 01:10:07.196105 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.789773
I0607 01:10:07.196116 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.835616
I0607 01:10:07.196130 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 1.52585 (* 1 = 1.52585 loss)
I0607 01:10:07.196143 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.686826 (* 1 = 0.686826 loss)
I0607 01:10:07.196157 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 1.10812 (* 0.0909091 = 0.100738 loss)
I0607 01:10:07.196171 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.908267 (* 0.0909091 = 0.0825697 loss)
I0607 01:10:07.196185 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 1.75171 (* 0.0909091 = 0.159246 loss)
I0607 01:10:07.196199 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 1.47795 (* 0.0909091 = 0.134359 loss)
I0607 01:10:07.196213 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 2.29853 (* 0.0909091 = 0.208957 loss)
I0607 01:10:07.196226 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 1.03801 (* 0.0909091 = 0.0943649 loss)
I0607 01:10:07.196240 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.550218 (* 0.0909091 = 0.0500198 loss)
I0607 01:10:07.196254 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 1.00837 (* 0.0909091 = 0.0916698 loss)
I0607 01:10:07.196269 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.441641 (* 0.0909091 = 0.0401492 loss)
I0607 01:10:07.196282 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.747492 (* 0.0909091 = 0.0679538 loss)
I0607 01:10:07.196296 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 1.23734 (* 0.0909091 = 0.112485 loss)
I0607 01:10:07.196310 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.853267 (* 0.0909091 = 0.0775697 loss)
I0607 01:10:07.196323 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.320959 (* 0.0909091 = 0.0291781 loss)
I0607 01:10:07.196337 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.340601 (* 0.0909091 = 0.0309637 loss)
I0607 01:10:07.196352 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 1.07146 (* 0.0909091 = 0.0974057 loss)
I0607 01:10:07.196365 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.38241 (* 0.0909091 = 0.0347645 loss)
I0607 01:10:07.196379 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.222642 (* 0.0909091 = 0.0202402 loss)
I0607 01:10:07.196393 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.348478 (* 0.0909091 = 0.0316798 loss)
I0607 01:10:07.196408 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.00533355 (* 0.0909091 = 0.000484868 loss)
I0607 01:10:07.196421 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.00225905 (* 0.0909091 = 0.000205368 loss)
I0607 01:10:07.196434 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.00173328 (* 0.0909091 = 0.000157571 loss)
I0607 01:10:07.196449 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 0.000438828 (* 0.0909091 = 3.98935e-05 loss)
I0607 01:10:07.196461 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.125
I0607 01:10:07.196473 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.125
I0607 01:10:07.196485 32403 solver.cpp:245] Train net output #149: total_confidence = 0.107395
I0607 01:10:07.196506 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.0668123
I0607 01:10:07.196521 32403 sgd_solver.cpp:106] Iteration 500, lr = 0.001
I0607 01:10:18.589634 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.1508 > 30) by scale factor 0.994999
I0607 01:10:56.255784 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.9098 > 30) by scale factor 0.940149
I0607 01:11:12.734740 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 57.1804 > 30) by scale factor 0.524655
I0607 01:11:13.521620 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 39.2868 > 30) by scale factor 0.763615
I0607 01:11:24.540010 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.3359 > 30) by scale factor 0.988927
I0607 01:11:37.081316 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 47.0777 > 30) by scale factor 0.637244
I0607 01:12:48.391294 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 42.4154 > 30) by scale factor 0.70729
I0607 01:13:36.902114 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.7657 > 30) by scale factor 0.944416
I0607 01:14:23.858458 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 36.1606 > 30) by scale factor 0.829632
I0607 01:14:49.755015 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.0316 > 30) by scale factor 0.966755
I0607 01:15:16.351179 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 47.0843 > 30) by scale factor 0.637155
I0607 01:16:06.477812 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.7501 > 30) by scale factor 0.944878
I0607 01:16:38.957581 32403 solver.cpp:229] Iteration 1000, loss = 4.52437
I0607 01:16:38.957748 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.383333
I0607 01:16:38.957770 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.5
I0607 01:16:38.957785 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.625
I0607 01:16:38.957798 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.25
I0607 01:16:38.957810 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.25
I0607 01:16:38.957823 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.25
I0607 01:16:38.957835 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.375
I0607 01:16:38.957847 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.5
I0607 01:16:38.957869 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 1
I0607 01:16:38.957882 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.75
I0607 01:16:38.957895 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.75
I0607 01:16:38.957907 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 1
I0607 01:16:38.957919 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 01:16:38.957931 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.875
I0607 01:16:38.957952 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.875
I0607 01:16:38.957964 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 0.875
I0607 01:16:38.957976 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 0.875
I0607 01:16:38.957988 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 0.875
I0607 01:16:38.958000 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 01:16:38.958012 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 01:16:38.958024 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 01:16:38.958037 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 01:16:38.958050 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 01:16:38.958061 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.732955
I0607 01:16:38.958073 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.65
I0607 01:16:38.958091 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 2.11361 (* 0.3 = 0.634083 loss)
I0607 01:16:38.958106 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.996329 (* 0.3 = 0.298899 loss)
I0607 01:16:38.958122 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 2.23579 (* 0.0272727 = 0.0609761 loss)
I0607 01:16:38.958135 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 2.32063 (* 0.0272727 = 0.06329 loss)
I0607 01:16:38.958149 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 3.07042 (* 0.0272727 = 0.0837386 loss)
I0607 01:16:38.958163 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 2.25641 (* 0.0272727 = 0.0615383 loss)
I0607 01:16:38.958178 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 2.38255 (* 0.0272727 = 0.0649787 loss)
I0607 01:16:38.958191 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 2.53882 (* 0.0272727 = 0.0692405 loss)
I0607 01:16:38.958205 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 1.56271 (* 0.0272727 = 0.0426193 loss)
I0607 01:16:38.958220 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.598586 (* 0.0272727 = 0.0163251 loss)
I0607 01:16:38.958235 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.936162 (* 0.0272727 = 0.0255317 loss)
I0607 01:16:38.958248 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.837982 (* 0.0272727 = 0.0228541 loss)
I0607 01:16:38.958262 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.559324 (* 0.0272727 = 0.0152543 loss)
I0607 01:16:38.958276 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.672115 (* 0.0272727 = 0.0183304 loss)
I0607 01:16:38.958313 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.449717 (* 0.0272727 = 0.012265 loss)
I0607 01:16:38.958329 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.48006 (* 0.0272727 = 0.0130925 loss)
I0607 01:16:38.958343 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.598039 (* 0.0272727 = 0.0163102 loss)
I0607 01:16:38.958358 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.483541 (* 0.0272727 = 0.0131875 loss)
I0607 01:16:38.958371 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.582876 (* 0.0272727 = 0.0158966 loss)
I0607 01:16:38.958386 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.0461028 (* 0.0272727 = 0.00125735 loss)
I0607 01:16:38.958401 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.0222725 (* 0.0272727 = 0.000607431 loss)
I0607 01:16:38.958415 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.0121534 (* 0.0272727 = 0.000331456 loss)
I0607 01:16:38.958430 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.00691503 (* 0.0272727 = 0.000188592 loss)
I0607 01:16:38.958444 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.00831424 (* 0.0272727 = 0.000226752 loss)
I0607 01:16:38.958463 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.466667
I0607 01:16:38.958475 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.75
I0607 01:16:38.958487 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.625
I0607 01:16:38.958500 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.5
I0607 01:16:38.958513 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.25
I0607 01:16:38.958531 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.25
I0607 01:16:38.958544 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.625
I0607 01:16:38.958555 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.375
I0607 01:16:38.958567 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.75
I0607 01:16:38.958580 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.75
I0607 01:16:38.958591 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 1
I0607 01:16:38.958602 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.75
I0607 01:16:38.958616 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 1
I0607 01:16:38.958627 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.875
I0607 01:16:38.958639 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.875
I0607 01:16:38.958652 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 0.875
I0607 01:16:38.958663 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 0.875
I0607 01:16:38.958675 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 0.875
I0607 01:16:38.958688 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 01:16:38.958698 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 01:16:38.958710 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 01:16:38.958722 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 01:16:38.958734 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 01:16:38.958745 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.761364
I0607 01:16:38.958762 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.733333
I0607 01:16:38.958776 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 2.00682 (* 0.3 = 0.602045 loss)
I0607 01:16:38.958791 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.92608 (* 0.3 = 0.277824 loss)
I0607 01:16:38.958806 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 1.64419 (* 0.0272727 = 0.0448414 loss)
I0607 01:16:38.958819 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 1.35532 (* 0.0272727 = 0.0369633 loss)
I0607 01:16:38.958845 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 2.16481 (* 0.0272727 = 0.0590403 loss)
I0607 01:16:38.958860 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.95861 (* 0.0272727 = 0.0534167 loss)
I0607 01:16:38.958874 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 2.45375 (* 0.0272727 = 0.0669206 loss)
I0607 01:16:38.958889 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.84026 (* 0.0272727 = 0.0501889 loss)
I0607 01:16:38.958902 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 1.80494 (* 0.0272727 = 0.0492255 loss)
I0607 01:16:38.958917 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.764812 (* 0.0272727 = 0.0208585 loss)
I0607 01:16:38.958933 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.913865 (* 0.0272727 = 0.0249236 loss)
I0607 01:16:38.958947 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.44733 (* 0.0272727 = 0.0121999 loss)
I0607 01:16:38.958961 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.52785 (* 0.0272727 = 0.0143959 loss)
I0607 01:16:38.958976 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.285232 (* 0.0272727 = 0.00777906 loss)
I0607 01:16:38.958991 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.343663 (* 0.0272727 = 0.00937264 loss)
I0607 01:16:38.959005 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.516098 (* 0.0272727 = 0.0140754 loss)
I0607 01:16:38.959019 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.616141 (* 0.0272727 = 0.0168038 loss)
I0607 01:16:38.959033 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.473196 (* 0.0272727 = 0.0129054 loss)
I0607 01:16:38.959048 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.512942 (* 0.0272727 = 0.0139893 loss)
I0607 01:16:38.959062 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.0119067 (* 0.0272727 = 0.000324729 loss)
I0607 01:16:38.959075 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.0051286 (* 0.0272727 = 0.000139871 loss)
I0607 01:16:38.959086 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.00231521 (* 0.0272727 = 6.31421e-05 loss)
I0607 01:16:38.959096 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.00201719 (* 0.0272727 = 5.50142e-05 loss)
I0607 01:16:38.959111 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 0.00111024 (* 0.0272727 = 3.02794e-05 loss)
I0607 01:16:38.959123 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.65
I0607 01:16:38.959136 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 0.875
I0607 01:16:38.959147 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.75
I0607 01:16:38.959159 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.625
I0607 01:16:38.959172 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.625
I0607 01:16:38.959183 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.5
I0607 01:16:38.959194 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.5
I0607 01:16:38.959206 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.75
I0607 01:16:38.959218 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.875
I0607 01:16:38.959230 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.75
I0607 01:16:38.959242 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.75
I0607 01:16:38.959254 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.75
I0607 01:16:38.959265 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.875
I0607 01:16:38.959277 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 0.875
I0607 01:16:38.959290 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 0.875
I0607 01:16:38.959300 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 0.875
I0607 01:16:38.959312 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 0.875
I0607 01:16:38.959336 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 0.875
I0607 01:16:38.959348 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 01:16:38.959360 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 01:16:38.959372 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 01:16:38.959384 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 01:16:38.959396 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 01:16:38.959408 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.823864
I0607 01:16:38.959420 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.85
I0607 01:16:38.959434 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 1.13857 (* 1 = 1.13857 loss)
I0607 01:16:38.959449 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.608847 (* 1 = 0.608847 loss)
I0607 01:16:38.959462 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 1.11526 (* 0.0909091 = 0.101388 loss)
I0607 01:16:38.959476 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 1.09525 (* 0.0909091 = 0.099568 loss)
I0607 01:16:38.959491 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 1.73588 (* 0.0909091 = 0.157807 loss)
I0607 01:16:38.959503 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.986957 (* 0.0909091 = 0.0897233 loss)
I0607 01:16:38.959517 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 1.67951 (* 0.0909091 = 0.152683 loss)
I0607 01:16:38.959532 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 1.58179 (* 0.0909091 = 0.143799 loss)
I0607 01:16:38.959544 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.593204 (* 0.0909091 = 0.0539276 loss)
I0607 01:16:38.959559 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.427236 (* 0.0909091 = 0.0388396 loss)
I0607 01:16:38.959573 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.63915 (* 0.0909091 = 0.0581045 loss)
I0607 01:16:38.959586 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.59223 (* 0.0909091 = 0.0538391 loss)
I0607 01:16:38.959600 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.556395 (* 0.0909091 = 0.0505813 loss)
I0607 01:16:38.959614 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.431322 (* 0.0909091 = 0.0392111 loss)
I0607 01:16:38.959627 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.351673 (* 0.0909091 = 0.0319702 loss)
I0607 01:16:38.959641 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.28584 (* 0.0909091 = 0.0259855 loss)
I0607 01:16:38.959656 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.385476 (* 0.0909091 = 0.0350433 loss)
I0607 01:16:38.959669 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.423957 (* 0.0909091 = 0.0385416 loss)
I0607 01:16:38.959683 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.321563 (* 0.0909091 = 0.029233 loss)
I0607 01:16:38.959697 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.0129756 (* 0.0909091 = 0.0011796 loss)
I0607 01:16:38.959712 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.00728195 (* 0.0909091 = 0.000661995 loss)
I0607 01:16:38.959725 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.00345143 (* 0.0909091 = 0.000313766 loss)
I0607 01:16:38.959739 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.00198684 (* 0.0909091 = 0.000180622 loss)
I0607 01:16:38.959753 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 0.000866708 (* 0.0909091 = 7.87916e-05 loss)
I0607 01:16:38.959766 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.25
I0607 01:16:38.959779 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.25
I0607 01:16:38.959790 32403 solver.cpp:245] Train net output #149: total_confidence = 0.166048
I0607 01:16:38.959815 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.12356
I0607 01:16:38.959830 32403 sgd_solver.cpp:106] Iteration 1000, lr = 0.001
I0607 01:17:22.308600 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.8133 > 30) by scale factor 0.943001
I0607 01:18:49.757988 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.7898 > 30) by scale factor 0.887842
I0607 01:18:54.444154 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.8599 > 30) by scale factor 0.972135
I0607 01:20:00.853502 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.0381 > 30) by scale factor 0.881364
I0607 01:20:03.198454 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.1184 > 30) by scale factor 0.905842
I0607 01:21:49.242472 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 35.5236 > 30) by scale factor 0.84451
I0607 01:22:49.323377 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 36.85 > 30) by scale factor 0.814111
I0607 01:23:09.229611 32403 solver.cpp:229] Iteration 1500, loss = 4.38283
I0607 01:23:09.229679 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.416667
I0607 01:23:09.229698 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.75
I0607 01:23:09.229712 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.75
I0607 01:23:09.229724 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.375
I0607 01:23:09.229739 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.125
I0607 01:23:09.229753 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.375
I0607 01:23:09.229764 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.75
I0607 01:23:09.229778 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.5
I0607 01:23:09.229789 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.875
I0607 01:23:09.229802 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.75
I0607 01:23:09.229815 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.875
I0607 01:23:09.229827 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.875
I0607 01:23:09.229840 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 01:23:09.229851 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 1
I0607 01:23:09.229864 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.875
I0607 01:23:09.229876 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 0.875
I0607 01:23:09.229888 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 01:23:09.229903 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 01:23:09.229915 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 01:23:09.229928 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 01:23:09.229939 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 01:23:09.229951 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 01:23:09.229964 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 01:23:09.229976 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.789773
I0607 01:23:09.229992 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.65
I0607 01:23:09.230008 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.84489 (* 0.3 = 0.553468 loss)
I0607 01:23:09.230023 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.682282 (* 0.3 = 0.204685 loss)
I0607 01:23:09.230038 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 1.13799 (* 0.0272727 = 0.0310362 loss)
I0607 01:23:09.230052 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.23514 (* 0.0272727 = 0.0336857 loss)
I0607 01:23:09.230067 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 2.61275 (* 0.0272727 = 0.0712568 loss)
I0607 01:23:09.230080 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 2.46398 (* 0.0272727 = 0.0671993 loss)
I0607 01:23:09.230094 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.89811 (* 0.0272727 = 0.0517665 loss)
I0607 01:23:09.230108 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.34428 (* 0.0272727 = 0.0366622 loss)
I0607 01:23:09.230123 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 1.62128 (* 0.0272727 = 0.0442168 loss)
I0607 01:23:09.230136 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.527337 (* 0.0272727 = 0.0143819 loss)
I0607 01:23:09.230150 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.71017 (* 0.0272727 = 0.0193683 loss)
I0607 01:23:09.230165 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.729183 (* 0.0272727 = 0.0198868 loss)
I0607 01:23:09.230178 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.5664 (* 0.0272727 = 0.0154473 loss)
I0607 01:23:09.230193 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.446448 (* 0.0272727 = 0.0121759 loss)
I0607 01:23:09.230252 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.211592 (* 0.0272727 = 0.0057707 loss)
I0607 01:23:09.230268 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.316877 (* 0.0272727 = 0.0086421 loss)
I0607 01:23:09.230281 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.397329 (* 0.0272727 = 0.0108363 loss)
I0607 01:23:09.230296 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.0905041 (* 0.0272727 = 0.00246829 loss)
I0607 01:23:09.230310 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.00301852 (* 0.0272727 = 8.23234e-05 loss)
I0607 01:23:09.230325 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.00037734 (* 0.0272727 = 1.02911e-05 loss)
I0607 01:23:09.230340 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.000171288 (* 0.0272727 = 4.67149e-06 loss)
I0607 01:23:09.230353 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.000134292 (* 0.0272727 = 3.66252e-06 loss)
I0607 01:23:09.230367 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.000161551 (* 0.0272727 = 4.40593e-06 loss)
I0607 01:23:09.230382 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.000142181 (* 0.0272727 = 3.87766e-06 loss)
I0607 01:23:09.230394 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.7
I0607 01:23:09.230407 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.875
I0607 01:23:09.230419 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.75
I0607 01:23:09.230432 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.5
I0607 01:23:09.230443 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.5
I0607 01:23:09.230455 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.375
I0607 01:23:09.230468 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.625
I0607 01:23:09.230479 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.625
I0607 01:23:09.230491 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.875
I0607 01:23:09.230504 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.75
I0607 01:23:09.230515 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.875
I0607 01:23:09.230527 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.875
I0607 01:23:09.230540 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.875
I0607 01:23:09.230551 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.875
I0607 01:23:09.230563 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.875
I0607 01:23:09.230576 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 0.875
I0607 01:23:09.230588 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 01:23:09.230600 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 01:23:09.230612 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 01:23:09.230623 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 01:23:09.230635 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 01:23:09.230646 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 01:23:09.230659 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 01:23:09.230670 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.880682
I0607 01:23:09.230681 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.833333
I0607 01:23:09.230696 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.17252 (* 0.3 = 0.351757 loss)
I0607 01:23:09.230710 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.453868 (* 0.3 = 0.13616 loss)
I0607 01:23:09.230723 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.502235 (* 0.0272727 = 0.0136973 loss)
I0607 01:23:09.230748 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.779341 (* 0.0272727 = 0.0212548 loss)
I0607 01:23:09.230763 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 1.61934 (* 0.0272727 = 0.0441638 loss)
I0607 01:23:09.230777 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 2.1547 (* 0.0272727 = 0.0587645 loss)
I0607 01:23:09.230794 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.68524 (* 0.0272727 = 0.045961 loss)
I0607 01:23:09.230809 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 0.934163 (* 0.0272727 = 0.0254772 loss)
I0607 01:23:09.230823 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 1.08218 (* 0.0272727 = 0.0295141 loss)
I0607 01:23:09.230837 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.541509 (* 0.0272727 = 0.0147684 loss)
I0607 01:23:09.230851 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.79545 (* 0.0272727 = 0.0216941 loss)
I0607 01:23:09.230865 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.503182 (* 0.0272727 = 0.0137232 loss)
I0607 01:23:09.230880 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.400854 (* 0.0272727 = 0.0109324 loss)
I0607 01:23:09.230893 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.420483 (* 0.0272727 = 0.0114677 loss)
I0607 01:23:09.230907 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.330743 (* 0.0272727 = 0.00902028 loss)
I0607 01:23:09.230921 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.311297 (* 0.0272727 = 0.00848993 loss)
I0607 01:23:09.230937 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.335027 (* 0.0272727 = 0.0091371 loss)
I0607 01:23:09.230953 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.0736344 (* 0.0272727 = 0.00200821 loss)
I0607 01:23:09.230968 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.00466257 (* 0.0272727 = 0.000127161 loss)
I0607 01:23:09.230983 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.000684146 (* 0.0272727 = 1.86585e-05 loss)
I0607 01:23:09.230998 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.000390115 (* 0.0272727 = 1.06395e-05 loss)
I0607 01:23:09.231011 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.000288972 (* 0.0272727 = 7.88105e-06 loss)
I0607 01:23:09.231025 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.000155595 (* 0.0272727 = 4.2435e-06 loss)
I0607 01:23:09.231040 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 8.39392e-05 (* 0.0272727 = 2.28925e-06 loss)
I0607 01:23:09.231052 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.816667
I0607 01:23:09.231065 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 01:23:09.231076 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.875
I0607 01:23:09.231088 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.875
I0607 01:23:09.231099 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.75
I0607 01:23:09.231112 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.875
I0607 01:23:09.231123 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.875
I0607 01:23:09.231135 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.625
I0607 01:23:09.231147 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 1
I0607 01:23:09.231158 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.75
I0607 01:23:09.231170 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.875
I0607 01:23:09.231181 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.875
I0607 01:23:09.231194 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 1
I0607 01:23:09.231204 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 0.875
I0607 01:23:09.231216 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 01:23:09.231227 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 01:23:09.231250 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 01:23:09.231263 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 01:23:09.231276 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 01:23:09.231287 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 01:23:09.231299 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 01:23:09.231312 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 01:23:09.231323 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 01:23:09.231334 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.9375
I0607 01:23:09.231346 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.966667
I0607 01:23:09.231360 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.530062 (* 1 = 0.530062 loss)
I0607 01:23:09.231374 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.190823 (* 1 = 0.190823 loss)
I0607 01:23:09.231389 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.0604207 (* 0.0909091 = 0.0054928 loss)
I0607 01:23:09.231403 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.421011 (* 0.0909091 = 0.0382737 loss)
I0607 01:23:09.231417 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.484814 (* 0.0909091 = 0.044074 loss)
I0607 01:23:09.231431 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.951086 (* 0.0909091 = 0.0864624 loss)
I0607 01:23:09.231446 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.403562 (* 0.0909091 = 0.0366875 loss)
I0607 01:23:09.231459 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.587071 (* 0.0909091 = 0.0533701 loss)
I0607 01:23:09.231473 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.990968 (* 0.0909091 = 0.090088 loss)
I0607 01:23:09.231487 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.111018 (* 0.0909091 = 0.0100926 loss)
I0607 01:23:09.231501 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.461372 (* 0.0909091 = 0.0419429 loss)
I0607 01:23:09.231516 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.430648 (* 0.0909091 = 0.0391498 loss)
I0607 01:23:09.231530 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.329043 (* 0.0909091 = 0.029913 loss)
I0607 01:23:09.231544 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.125362 (* 0.0909091 = 0.0113965 loss)
I0607 01:23:09.231559 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.192732 (* 0.0909091 = 0.0175211 loss)
I0607 01:23:09.231572 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.172135 (* 0.0909091 = 0.0156486 loss)
I0607 01:23:09.231585 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.144448 (* 0.0909091 = 0.0131317 loss)
I0607 01:23:09.231600 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.0837439 (* 0.0909091 = 0.00761308 loss)
I0607 01:23:09.231613 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.00404558 (* 0.0909091 = 0.00036778 loss)
I0607 01:23:09.231627 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.000594013 (* 0.0909091 = 5.40012e-05 loss)
I0607 01:23:09.231640 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.000251033 (* 0.0909091 = 2.28212e-05 loss)
I0607 01:23:09.231654 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.000261516 (* 0.0909091 = 2.37742e-05 loss)
I0607 01:23:09.231668 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.000222563 (* 0.0909091 = 2.0233e-05 loss)
I0607 01:23:09.231683 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 0.000136472 (* 0.0909091 = 1.24066e-05 loss)
I0607 01:23:09.231694 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.375
I0607 01:23:09.231706 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.375
I0607 01:23:09.231729 32403 solver.cpp:245] Train net output #149: total_confidence = 0.335031
I0607 01:23:09.231741 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.265859
I0607 01:23:09.231755 32403 sgd_solver.cpp:106] Iteration 1500, lr = 0.001
I0607 01:23:11.170199 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 35.8079 > 30) by scale factor 0.837805
I0607 01:23:13.510813 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.2164 > 30) by scale factor 0.806097
I0607 01:26:59.692255 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.5715 > 30) by scale factor 0.798477
I0607 01:27:50.438215 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 40.8913 > 30) by scale factor 0.733653
I0607 01:28:09.139561 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 38.9265 > 30) by scale factor 0.770683
I0607 01:28:19.268362 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.6021 > 30) by scale factor 0.949304
I0607 01:28:37.196938 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.9435 > 30) by scale factor 0.939158
I0607 01:29:39.239522 32403 solver.cpp:229] Iteration 2000, loss = 4.16339
I0607 01:29:39.239657 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.428571
I0607 01:29:39.239688 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 1
I0607 01:29:39.239711 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.5
I0607 01:29:39.239734 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.625
I0607 01:29:39.239756 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.5
I0607 01:29:39.239778 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.75
I0607 01:29:39.239797 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.25
I0607 01:29:39.239820 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.75
I0607 01:29:39.239841 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.75
I0607 01:29:39.239861 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.625
I0607 01:29:39.239884 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.75
I0607 01:29:39.239905 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.875
I0607 01:29:39.239925 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 01:29:39.239946 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.875
I0607 01:29:39.239966 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.875
I0607 01:29:39.239986 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 0.875
I0607 01:29:39.240010 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 0.875
I0607 01:29:39.240032 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 01:29:39.240052 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 01:29:39.240072 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 01:29:39.240092 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 01:29:39.240113 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 01:29:39.240134 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 01:29:39.240154 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.784091
I0607 01:29:39.240175 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.68254
I0607 01:29:39.240200 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.87989 (* 0.3 = 0.563968 loss)
I0607 01:29:39.240226 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.700379 (* 0.3 = 0.210114 loss)
I0607 01:29:39.240252 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.344799 (* 0.0272727 = 0.00940361 loss)
I0607 01:29:39.240279 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.81901 (* 0.0272727 = 0.0496093 loss)
I0607 01:29:39.240305 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.47557 (* 0.0272727 = 0.0402429 loss)
I0607 01:29:39.240330 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.70934 (* 0.0272727 = 0.0466183 loss)
I0607 01:29:39.240355 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.30418 (* 0.0272727 = 0.0355685 loss)
I0607 01:29:39.240381 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 2.11596 (* 0.0272727 = 0.0577079 loss)
I0607 01:29:39.240406 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 1.12156 (* 0.0272727 = 0.0305881 loss)
I0607 01:29:39.240430 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 1.35907 (* 0.0272727 = 0.0370654 loss)
I0607 01:29:39.240454 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 1.43569 (* 0.0272727 = 0.0391553 loss)
I0607 01:29:39.240479 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 1.26044 (* 0.0272727 = 0.0343757 loss)
I0607 01:29:39.240504 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.452528 (* 0.0272727 = 0.0123417 loss)
I0607 01:29:39.240528 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.466772 (* 0.0272727 = 0.0127301 loss)
I0607 01:29:39.240586 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.385018 (* 0.0272727 = 0.0105005 loss)
I0607 01:29:39.240612 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.495245 (* 0.0272727 = 0.0135067 loss)
I0607 01:29:39.240638 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.524372 (* 0.0272727 = 0.0143011 loss)
I0607 01:29:39.240661 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.27662 (* 0.0272727 = 0.00754419 loss)
I0607 01:29:39.240692 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.0138106 (* 0.0272727 = 0.000376653 loss)
I0607 01:29:39.240720 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.00483906 (* 0.0272727 = 0.000131974 loss)
I0607 01:29:39.240747 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.00275553 (* 0.0272727 = 7.51507e-05 loss)
I0607 01:29:39.240772 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.00153716 (* 0.0272727 = 4.19226e-05 loss)
I0607 01:29:39.240797 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.00116414 (* 0.0272727 = 3.17493e-05 loss)
I0607 01:29:39.240821 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.000440536 (* 0.0272727 = 1.20146e-05 loss)
I0607 01:29:39.240842 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.603175
I0607 01:29:39.240864 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 1
I0607 01:29:39.240885 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.875
I0607 01:29:39.240905 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.875
I0607 01:29:39.240927 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.75
I0607 01:29:39.240949 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.75
I0607 01:29:39.240970 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.5
I0607 01:29:39.240990 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.75
I0607 01:29:39.241011 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.875
I0607 01:29:39.241030 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.75
I0607 01:29:39.241050 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.875
I0607 01:29:39.241070 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.75
I0607 01:29:39.241091 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.875
I0607 01:29:39.241111 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.875
I0607 01:29:39.241166 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.875
I0607 01:29:39.241199 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 0.875
I0607 01:29:39.241221 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 01:29:39.241241 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 01:29:39.241262 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 01:29:39.241284 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 01:29:39.241307 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 01:29:39.241328 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 01:29:39.241348 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 01:29:39.241369 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.857955
I0607 01:29:39.241389 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.793651
I0607 01:29:39.241415 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.46224 (* 0.3 = 0.438673 loss)
I0607 01:29:39.241439 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.531425 (* 0.3 = 0.159427 loss)
I0607 01:29:39.241466 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.10118 (* 0.0272727 = 0.00275946 loss)
I0607 01:29:39.241492 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.613184 (* 0.0272727 = 0.0167232 loss)
I0607 01:29:39.241534 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 1.01538 (* 0.0272727 = 0.027692 loss)
I0607 01:29:39.241562 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.30053 (* 0.0272727 = 0.0354691 loss)
I0607 01:29:39.241587 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 0.760057 (* 0.0272727 = 0.0207288 loss)
I0607 01:29:39.241613 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.1639 (* 0.0272727 = 0.0317426 loss)
I0607 01:29:39.241638 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.960119 (* 0.0272727 = 0.0261851 loss)
I0607 01:29:39.241663 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.683679 (* 0.0272727 = 0.0186458 loss)
I0607 01:29:39.241689 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 1.19409 (* 0.0272727 = 0.0325661 loss)
I0607 01:29:39.241714 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.88023 (* 0.0272727 = 0.0240063 loss)
I0607 01:29:39.241744 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.574512 (* 0.0272727 = 0.0156685 loss)
I0607 01:29:39.241770 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.381221 (* 0.0272727 = 0.0103969 loss)
I0607 01:29:39.241796 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.289445 (* 0.0272727 = 0.00789395 loss)
I0607 01:29:39.241822 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.506569 (* 0.0272727 = 0.0138155 loss)
I0607 01:29:39.241848 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.442215 (* 0.0272727 = 0.0120604 loss)
I0607 01:29:39.241873 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.18787 (* 0.0272727 = 0.00512372 loss)
I0607 01:29:39.241899 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.00101773 (* 0.0272727 = 2.77563e-05 loss)
I0607 01:29:39.241925 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.000197689 (* 0.0272727 = 5.39151e-06 loss)
I0607 01:29:39.241951 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 3.90516e-05 (* 0.0272727 = 1.06504e-06 loss)
I0607 01:29:39.241979 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 6.62011e-05 (* 0.0272727 = 1.80548e-06 loss)
I0607 01:29:39.242008 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 6.05463e-05 (* 0.0272727 = 1.65126e-06 loss)
I0607 01:29:39.242035 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 2.35449e-05 (* 0.0272727 = 6.42134e-07 loss)
I0607 01:29:39.242058 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.777778
I0607 01:29:39.242079 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 01:29:39.242100 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.875
I0607 01:29:39.242121 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 1
I0607 01:29:39.242142 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.875
I0607 01:29:39.242164 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.875
I0607 01:29:39.242184 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.75
I0607 01:29:39.242205 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.875
I0607 01:29:39.242228 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.875
I0607 01:29:39.242247 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.75
I0607 01:29:39.242267 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.875
I0607 01:29:39.242288 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.75
I0607 01:29:39.242308 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 1
I0607 01:29:39.242329 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 0.875
I0607 01:29:39.242349 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 0.875
I0607 01:29:39.242369 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 0.875
I0607 01:29:39.242405 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 0.875
I0607 01:29:39.242434 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 01:29:39.242455 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 01:29:39.242473 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 01:29:39.242491 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 01:29:39.242509 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 01:29:39.242528 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 01:29:39.242550 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.920455
I0607 01:29:39.242571 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.873016
I0607 01:29:39.242596 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.902617 (* 1 = 0.902617 loss)
I0607 01:29:39.242621 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.328922 (* 1 = 0.328922 loss)
I0607 01:29:39.242647 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.00923907 (* 0.0909091 = 0.000839915 loss)
I0607 01:29:39.242673 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.308648 (* 0.0909091 = 0.0280589 loss)
I0607 01:29:39.242699 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.143352 (* 0.0909091 = 0.013032 loss)
I0607 01:29:39.242723 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.55597 (* 0.0909091 = 0.0505428 loss)
I0607 01:29:39.242748 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.37418 (* 0.0909091 = 0.0340164 loss)
I0607 01:29:39.242779 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.866057 (* 0.0909091 = 0.0787324 loss)
I0607 01:29:39.242805 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.552953 (* 0.0909091 = 0.0502684 loss)
I0607 01:29:39.242830 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.93327 (* 0.0909091 = 0.0848427 loss)
I0607 01:29:39.242856 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.897753 (* 0.0909091 = 0.0816139 loss)
I0607 01:29:39.242880 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.341887 (* 0.0909091 = 0.0310807 loss)
I0607 01:29:39.242905 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.441049 (* 0.0909091 = 0.0400953 loss)
I0607 01:29:39.242930 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.230292 (* 0.0909091 = 0.0209357 loss)
I0607 01:29:39.242954 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.229502 (* 0.0909091 = 0.0208638 loss)
I0607 01:29:39.242980 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.641255 (* 0.0909091 = 0.0582959 loss)
I0607 01:29:39.243003 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.583393 (* 0.0909091 = 0.0530358 loss)
I0607 01:29:39.243031 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.267779 (* 0.0909091 = 0.0243436 loss)
I0607 01:29:39.243058 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.0011659 (* 0.0909091 = 0.00010599 loss)
I0607 01:29:39.243079 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.000203336 (* 0.0909091 = 1.84851e-05 loss)
I0607 01:29:39.243108 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.00011643 (* 0.0909091 = 1.05846e-05 loss)
I0607 01:29:39.243134 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 5.3611e-05 (* 0.0909091 = 4.87373e-06 loss)
I0607 01:29:39.243158 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 5.83307e-05 (* 0.0909091 = 5.30279e-06 loss)
I0607 01:29:39.243185 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 1.61681e-05 (* 0.0909091 = 1.46983e-06 loss)
I0607 01:29:39.243206 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.75
I0607 01:29:39.243228 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.625
I0607 01:29:39.243265 32403 solver.cpp:245] Train net output #149: total_confidence = 0.543452
I0607 01:29:39.243288 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.513039
I0607 01:29:39.243310 32403 sgd_solver.cpp:106] Iteration 2000, lr = 0.001
I0607 01:29:59.877012 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.5952 > 30) by scale factor 0.892984
I0607 01:30:17.054630 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.0784 > 30) by scale factor 0.9653
I0607 01:30:20.950425 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 54.4171 > 30) by scale factor 0.551297
I0607 01:30:52.116107 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 35.3265 > 30) by scale factor 0.849221
I0607 01:31:03.001390 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.4998 > 30) by scale factor 0.952387
I0607 01:32:05.974678 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.6278 > 30) by scale factor 0.866357
I0607 01:32:23.858770 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 35.7122 > 30) by scale factor 0.840049
I0607 01:32:28.543617 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.6498 > 30) by scale factor 0.891536
I0607 01:33:32.322119 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 43.6944 > 30) by scale factor 0.686587
I0607 01:34:12.797992 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 44.5904 > 30) by scale factor 0.67279
I0607 01:34:22.131364 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 41.0461 > 30) by scale factor 0.730885
I0607 01:35:26.714614 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 40.2455 > 30) by scale factor 0.745425
I0607 01:36:08.442008 32403 solver.cpp:229] Iteration 2500, loss = 4.19057
I0607 01:36:08.442162 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.403226
I0607 01:36:08.442183 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.875
I0607 01:36:08.442198 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.875
I0607 01:36:08.442209 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.375
I0607 01:36:08.442222 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.25
I0607 01:36:08.442234 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.5
I0607 01:36:08.442246 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.625
I0607 01:36:08.442258 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.625
I0607 01:36:08.442270 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.75
I0607 01:36:08.442282 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.875
I0607 01:36:08.442294 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.75
I0607 01:36:08.442308 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.75
I0607 01:36:08.442320 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.75
I0607 01:36:08.442332 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.875
I0607 01:36:08.442344 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.875
I0607 01:36:08.442356 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 01:36:08.442368 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 0.875
I0607 01:36:08.442380 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 0.875
I0607 01:36:08.442392 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 0.875
I0607 01:36:08.442404 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 0.875
I0607 01:36:08.442416 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 01:36:08.442428 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 01:36:08.442440 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 01:36:08.442452 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.772727
I0607 01:36:08.442464 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.709677
I0607 01:36:08.442481 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.88559 (* 0.3 = 0.565676 loss)
I0607 01:36:08.442495 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.753475 (* 0.3 = 0.226043 loss)
I0607 01:36:08.442510 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.763514 (* 0.0272727 = 0.0208231 loss)
I0607 01:36:08.442524 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 0.91419 (* 0.0272727 = 0.0249325 loss)
I0607 01:36:08.442538 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.36183 (* 0.0272727 = 0.0371407 loss)
I0607 01:36:08.442553 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.59882 (* 0.0272727 = 0.0436043 loss)
I0607 01:36:08.442566 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.35418 (* 0.0272727 = 0.0369322 loss)
I0607 01:36:08.442580 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 0.982106 (* 0.0272727 = 0.0267847 loss)
I0607 01:36:08.442595 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 0.99687 (* 0.0272727 = 0.0271874 loss)
I0607 01:36:08.442608 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.899562 (* 0.0272727 = 0.0245335 loss)
I0607 01:36:08.442622 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.665635 (* 0.0272727 = 0.0181537 loss)
I0607 01:36:08.442636 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 1.15047 (* 0.0272727 = 0.0313764 loss)
I0607 01:36:08.442651 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 1.1335 (* 0.0272727 = 0.0309137 loss)
I0607 01:36:08.442664 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 1.00797 (* 0.0272727 = 0.0274901 loss)
I0607 01:36:08.442705 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.75667 (* 0.0272727 = 0.0206365 loss)
I0607 01:36:08.442721 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.483877 (* 0.0272727 = 0.0131967 loss)
I0607 01:36:08.442735 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.227623 (* 0.0272727 = 0.0062079 loss)
I0607 01:36:08.442750 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.326476 (* 0.0272727 = 0.00890388 loss)
I0607 01:36:08.442764 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.493619 (* 0.0272727 = 0.0134623 loss)
I0607 01:36:08.442780 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.88597 (* 0.0272727 = 0.0241628 loss)
I0607 01:36:08.442793 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 1.23159 (* 0.0272727 = 0.0335888 loss)
I0607 01:36:08.442808 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.00054356 (* 0.0272727 = 1.48244e-05 loss)
I0607 01:36:08.442822 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.000381899 (* 0.0272727 = 1.04154e-05 loss)
I0607 01:36:08.442837 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.000293399 (* 0.0272727 = 8.00179e-06 loss)
I0607 01:36:08.442849 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.629032
I0607 01:36:08.442862 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.875
I0607 01:36:08.442876 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 1
I0607 01:36:08.442889 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.875
I0607 01:36:08.442901 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.75
I0607 01:36:08.442914 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.75
I0607 01:36:08.442925 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.875
I0607 01:36:08.442937 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.75
I0607 01:36:08.442950 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.75
I0607 01:36:08.442961 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.75
I0607 01:36:08.442973 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.75
I0607 01:36:08.442986 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.625
I0607 01:36:08.442997 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.75
I0607 01:36:08.443009 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.875
I0607 01:36:08.443022 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.875
I0607 01:36:08.443033 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 0.875
I0607 01:36:08.443045 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 0.875
I0607 01:36:08.443058 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 0.875
I0607 01:36:08.443069 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 0.875
I0607 01:36:08.443081 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 0.875
I0607 01:36:08.443094 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 01:36:08.443105 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 01:36:08.443117 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 01:36:08.443130 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.863636
I0607 01:36:08.443137 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.806452
I0607 01:36:08.443147 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.49549 (* 0.3 = 0.448646 loss)
I0607 01:36:08.443157 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.547175 (* 0.3 = 0.164152 loss)
I0607 01:36:08.443177 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.13996 (* 0.0272727 = 0.00381709 loss)
I0607 01:36:08.443192 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.159219 (* 0.0272727 = 0.00434233 loss)
I0607 01:36:08.443218 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.84552 (* 0.0272727 = 0.0230596 loss)
I0607 01:36:08.443233 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 0.721214 (* 0.0272727 = 0.0196695 loss)
I0607 01:36:08.443248 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.0369 (* 0.0272727 = 0.0282791 loss)
I0607 01:36:08.443261 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 0.495453 (* 0.0272727 = 0.0135124 loss)
I0607 01:36:08.443276 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.758981 (* 0.0272727 = 0.0206995 loss)
I0607 01:36:08.443290 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.754562 (* 0.0272727 = 0.020579 loss)
I0607 01:36:08.443303 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.676081 (* 0.0272727 = 0.0184386 loss)
I0607 01:36:08.443317 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.790452 (* 0.0272727 = 0.0215578 loss)
I0607 01:36:08.443331 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 1.04088 (* 0.0272727 = 0.0283877 loss)
I0607 01:36:08.443346 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.712737 (* 0.0272727 = 0.0194383 loss)
I0607 01:36:08.443359 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.516128 (* 0.0272727 = 0.0140762 loss)
I0607 01:36:08.443373 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.632149 (* 0.0272727 = 0.0172404 loss)
I0607 01:36:08.443388 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.27334 (* 0.0272727 = 0.00745471 loss)
I0607 01:36:08.443403 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.463034 (* 0.0272727 = 0.0126282 loss)
I0607 01:36:08.443416 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.978008 (* 0.0272727 = 0.0266729 loss)
I0607 01:36:08.443431 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 1.51616 (* 0.0272727 = 0.0413498 loss)
I0607 01:36:08.443445 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 1.994 (* 0.0272727 = 0.0543818 loss)
I0607 01:36:08.443459 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 4.85782e-06 (* 0.0272727 = 1.32486e-07 loss)
I0607 01:36:08.443473 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 2.02656e-06 (* 0.0272727 = 5.52699e-08 loss)
I0607 01:36:08.443488 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 2.75673e-06 (* 0.0272727 = 7.51834e-08 loss)
I0607 01:36:08.443500 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.709677
I0607 01:36:08.443512 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 0.875
I0607 01:36:08.443526 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 1
I0607 01:36:08.443537 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.875
I0607 01:36:08.443549 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 1
I0607 01:36:08.443560 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 1
I0607 01:36:08.443572 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.875
I0607 01:36:08.443584 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 1
I0607 01:36:08.443595 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.875
I0607 01:36:08.443608 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.75
I0607 01:36:08.443619 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.75
I0607 01:36:08.443630 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.75
I0607 01:36:08.443642 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.75
I0607 01:36:08.443655 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 0.875
I0607 01:36:08.443665 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 0.875
I0607 01:36:08.443677 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 0.875
I0607 01:36:08.443698 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 0.875
I0607 01:36:08.443712 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 0.875
I0607 01:36:08.443724 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 0.875
I0607 01:36:08.443737 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 0.875
I0607 01:36:08.443748 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 01:36:08.443760 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 01:36:08.443773 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 01:36:08.443783 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.897727
I0607 01:36:08.443796 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.83871
I0607 01:36:08.443810 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 1.31802 (* 1 = 1.31802 loss)
I0607 01:36:08.443825 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.46994 (* 1 = 0.46994 loss)
I0607 01:36:08.443840 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.522156 (* 0.0909091 = 0.0474688 loss)
I0607 01:36:08.443853 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.0265362 (* 0.0909091 = 0.00241238 loss)
I0607 01:36:08.443867 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.715152 (* 0.0909091 = 0.0650139 loss)
I0607 01:36:08.443881 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.230248 (* 0.0909091 = 0.0209316 loss)
I0607 01:36:08.443897 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.11191 (* 0.0909091 = 0.0101736 loss)
I0607 01:36:08.443907 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.498503 (* 0.0909091 = 0.0453184 loss)
I0607 01:36:08.443922 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.0589906 (* 0.0909091 = 0.00536279 loss)
I0607 01:36:08.443939 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.34795 (* 0.0909091 = 0.0316318 loss)
I0607 01:36:08.443953 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.715907 (* 0.0909091 = 0.0650824 loss)
I0607 01:36:08.443969 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 1.01362 (* 0.0909091 = 0.0921476 loss)
I0607 01:36:08.443982 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 1.24637 (* 0.0909091 = 0.113306 loss)
I0607 01:36:08.443996 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.68332 (* 0.0909091 = 0.06212 loss)
I0607 01:36:08.444010 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.721011 (* 0.0909091 = 0.0655464 loss)
I0607 01:36:08.444023 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.358637 (* 0.0909091 = 0.0326034 loss)
I0607 01:36:08.444037 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.312211 (* 0.0909091 = 0.0283828 loss)
I0607 01:36:08.444051 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.549269 (* 0.0909091 = 0.0499335 loss)
I0607 01:36:08.444066 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.663789 (* 0.0909091 = 0.0603445 loss)
I0607 01:36:08.444079 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.873225 (* 0.0909091 = 0.0793841 loss)
I0607 01:36:08.444093 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 1.35712 (* 0.0909091 = 0.123374 loss)
I0607 01:36:08.444108 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.000415333 (* 0.0909091 = 3.77575e-05 loss)
I0607 01:36:08.444121 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.000124869 (* 0.0909091 = 1.13517e-05 loss)
I0607 01:36:08.444136 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 5.17043e-05 (* 0.0909091 = 4.70039e-06 loss)
I0607 01:36:08.444149 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.5
I0607 01:36:08.444160 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.625
I0607 01:36:08.444181 32403 solver.cpp:245] Train net output #149: total_confidence = 0.46541
I0607 01:36:08.444195 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.472481
I0607 01:36:08.444207 32403 sgd_solver.cpp:106] Iteration 2500, lr = 0.001
I0607 01:36:10.394326 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.978 > 30) by scale factor 0.938146
I0607 01:38:18.690879 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 42.2745 > 30) by scale factor 0.709647
I0607 01:38:56.799852 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 35.8401 > 30) by scale factor 0.837051
I0607 01:39:03.002720 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 35.2194 > 30) by scale factor 0.851803
I0607 01:39:41.894419 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 41.9918 > 30) by scale factor 0.714426
I0607 01:40:28.501749 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 40.1724 > 30) by scale factor 0.746781
I0607 01:40:30.822355 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.1142 > 30) by scale factor 0.905955
I0607 01:41:36.108605 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.9926 > 30) by scale factor 0.882546
I0607 01:42:15.773216 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.7909 > 30) by scale factor 0.943668
I0607 01:42:37.131955 32403 solver.cpp:229] Iteration 3000, loss = 4.21465
I0607 01:42:37.132032 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.484848
I0607 01:42:37.132050 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.625
I0607 01:42:37.132064 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.375
I0607 01:42:37.132077 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.5
I0607 01:42:37.132089 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.5
I0607 01:42:37.132102 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.5
I0607 01:42:37.132113 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.5
I0607 01:42:37.132125 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.5
I0607 01:42:37.132138 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.75
I0607 01:42:37.132150 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.75
I0607 01:42:37.132163 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.875
I0607 01:42:37.132175 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.75
I0607 01:42:37.132187 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.75
I0607 01:42:37.132200 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.875
I0607 01:42:37.132211 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.75
I0607 01:42:37.132223 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 0.75
I0607 01:42:37.132236 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 0.875
I0607 01:42:37.132247 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 01:42:37.132261 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 01:42:37.132272 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 01:42:37.132284 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 01:42:37.132295 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 01:42:37.132308 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 01:42:37.132319 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.801136
I0607 01:42:37.132331 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.757576
I0607 01:42:37.132349 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.66752 (* 0.3 = 0.500257 loss)
I0607 01:42:37.132364 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.64457 (* 0.3 = 0.193371 loss)
I0607 01:42:37.132378 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 1.21407 (* 0.0272727 = 0.0331111 loss)
I0607 01:42:37.132392 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.47736 (* 0.0272727 = 0.0402916 loss)
I0607 01:42:37.132406 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.42291 (* 0.0272727 = 0.0388068 loss)
I0607 01:42:37.132419 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.72945 (* 0.0272727 = 0.0471668 loss)
I0607 01:42:37.132433 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.20916 (* 0.0272727 = 0.0329772 loss)
I0607 01:42:37.132447 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.60724 (* 0.0272727 = 0.0438337 loss)
I0607 01:42:37.132462 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 1.21642 (* 0.0272727 = 0.0331751 loss)
I0607 01:42:37.132475 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 1.011 (* 0.0272727 = 0.0275727 loss)
I0607 01:42:37.132490 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 1.41139 (* 0.0272727 = 0.0384926 loss)
I0607 01:42:37.132504 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.504907 (* 0.0272727 = 0.0137702 loss)
I0607 01:42:37.132519 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.529204 (* 0.0272727 = 0.0144328 loss)
I0607 01:42:37.132532 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.778537 (* 0.0272727 = 0.0212328 loss)
I0607 01:42:37.132594 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.700229 (* 0.0272727 = 0.0190972 loss)
I0607 01:42:37.132611 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.914577 (* 0.0272727 = 0.024943 loss)
I0607 01:42:37.132624 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.521743 (* 0.0272727 = 0.0142294 loss)
I0607 01:42:37.132638 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.377433 (* 0.0272727 = 0.0102936 loss)
I0607 01:42:37.132652 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.0111268 (* 0.0272727 = 0.000303458 loss)
I0607 01:42:37.132668 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.00118477 (* 0.0272727 = 3.23119e-05 loss)
I0607 01:42:37.132681 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.000310156 (* 0.0272727 = 8.45879e-06 loss)
I0607 01:42:37.132696 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.000249568 (* 0.0272727 = 6.8064e-06 loss)
I0607 01:42:37.132710 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 9.34182e-05 (* 0.0272727 = 2.54777e-06 loss)
I0607 01:42:37.132725 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 4.81584e-05 (* 0.0272727 = 1.31341e-06 loss)
I0607 01:42:37.132737 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.560606
I0607 01:42:37.132750 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.625
I0607 01:42:37.132766 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.625
I0607 01:42:37.132778 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.625
I0607 01:42:37.132791 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.75
I0607 01:42:37.132802 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.875
I0607 01:42:37.132814 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.625
I0607 01:42:37.132827 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.75
I0607 01:42:37.132838 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.75
I0607 01:42:37.132850 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.75
I0607 01:42:37.132863 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.875
I0607 01:42:37.132874 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.875
I0607 01:42:37.132886 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.75
I0607 01:42:37.132899 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 1
I0607 01:42:37.132910 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.75
I0607 01:42:37.132921 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 0.75
I0607 01:42:37.132933 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 0.875
I0607 01:42:37.132946 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 01:42:37.132957 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 01:42:37.132969 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 01:42:37.132982 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 01:42:37.132992 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 01:42:37.133004 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 01:42:37.133015 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.829545
I0607 01:42:37.133028 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.80303
I0607 01:42:37.133041 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.32497 (* 0.3 = 0.397491 loss)
I0607 01:42:37.133057 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.525575 (* 0.3 = 0.157672 loss)
I0607 01:42:37.133072 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.948133 (* 0.0272727 = 0.0258582 loss)
I0607 01:42:37.133098 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 1.03235 (* 0.0272727 = 0.0281551 loss)
I0607 01:42:37.133110 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 1.02687 (* 0.0272727 = 0.0280055 loss)
I0607 01:42:37.133132 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.0695 (* 0.0272727 = 0.0291681 loss)
I0607 01:42:37.133149 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 0.852631 (* 0.0272727 = 0.0232536 loss)
I0607 01:42:37.133163 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.14056 (* 0.0272727 = 0.0311063 loss)
I0607 01:42:37.133177 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 1.14726 (* 0.0272727 = 0.0312888 loss)
I0607 01:42:37.133191 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.972496 (* 0.0272727 = 0.0265226 loss)
I0607 01:42:37.133205 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.968747 (* 0.0272727 = 0.0264204 loss)
I0607 01:42:37.133219 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.488794 (* 0.0272727 = 0.0133307 loss)
I0607 01:42:37.133234 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.655748 (* 0.0272727 = 0.017884 loss)
I0607 01:42:37.133246 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.740528 (* 0.0272727 = 0.0201962 loss)
I0607 01:42:37.133260 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.46177 (* 0.0272727 = 0.0125937 loss)
I0607 01:42:37.133275 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.707787 (* 0.0272727 = 0.0193033 loss)
I0607 01:42:37.133288 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.531746 (* 0.0272727 = 0.0145022 loss)
I0607 01:42:37.133303 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.412469 (* 0.0272727 = 0.0112491 loss)
I0607 01:42:37.133317 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.0611727 (* 0.0272727 = 0.00166835 loss)
I0607 01:42:37.133332 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.0214332 (* 0.0272727 = 0.000584542 loss)
I0607 01:42:37.133345 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.00401789 (* 0.0272727 = 0.000109579 loss)
I0607 01:42:37.133359 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.00202011 (* 0.0272727 = 5.5094e-05 loss)
I0607 01:42:37.133373 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.00085929 (* 0.0272727 = 2.34352e-05 loss)
I0607 01:42:37.133388 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 0.000530683 (* 0.0272727 = 1.44732e-05 loss)
I0607 01:42:37.133400 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.681818
I0607 01:42:37.133412 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 0.875
I0607 01:42:37.133424 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 1
I0607 01:42:37.133436 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 1
I0607 01:42:37.133447 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 1
I0607 01:42:37.133460 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 1
I0607 01:42:37.133471 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.875
I0607 01:42:37.133482 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.75
I0607 01:42:37.133496 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.625
I0607 01:42:37.133507 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.625
I0607 01:42:37.133518 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.75
I0607 01:42:37.133530 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 1
I0607 01:42:37.133543 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.75
I0607 01:42:37.133553 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 0.875
I0607 01:42:37.133565 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 0.875
I0607 01:42:37.133576 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 01:42:37.133599 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 0.875
I0607 01:42:37.133613 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 01:42:37.133625 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 01:42:37.133636 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 01:42:37.133648 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 01:42:37.133661 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 01:42:37.133672 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 01:42:37.133683 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.880682
I0607 01:42:37.133697 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.893939
I0607 01:42:37.133710 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.879544 (* 1 = 0.879544 loss)
I0607 01:42:37.133724 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.35238 (* 1 = 0.35238 loss)
I0607 01:42:37.133738 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.333335 (* 0.0909091 = 0.0303032 loss)
I0607 01:42:37.133752 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.255195 (* 0.0909091 = 0.0231996 loss)
I0607 01:42:37.133766 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.315483 (* 0.0909091 = 0.0286803 loss)
I0607 01:42:37.133780 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.303748 (* 0.0909091 = 0.0276135 loss)
I0607 01:42:37.133795 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.124064 (* 0.0909091 = 0.0112786 loss)
I0607 01:42:37.133810 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.370007 (* 0.0909091 = 0.033637 loss)
I0607 01:42:37.133826 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.870303 (* 0.0909091 = 0.0791185 loss)
I0607 01:42:37.133839 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 1.14472 (* 0.0909091 = 0.104065 loss)
I0607 01:42:37.133855 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.811351 (* 0.0909091 = 0.0737592 loss)
I0607 01:42:37.133869 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.707958 (* 0.0909091 = 0.0643598 loss)
I0607 01:42:37.133883 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.318341 (* 0.0909091 = 0.0289401 loss)
I0607 01:42:37.133898 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.717368 (* 0.0909091 = 0.0652152 loss)
I0607 01:42:37.133911 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.425354 (* 0.0909091 = 0.0386686 loss)
I0607 01:42:37.133925 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.496866 (* 0.0909091 = 0.0451696 loss)
I0607 01:42:37.133939 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.326443 (* 0.0909091 = 0.0296766 loss)
I0607 01:42:37.133952 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.465136 (* 0.0909091 = 0.0422851 loss)
I0607 01:42:37.133966 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.0317731 (* 0.0909091 = 0.00288846 loss)
I0607 01:42:37.133980 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.0115959 (* 0.0909091 = 0.00105417 loss)
I0607 01:42:37.133994 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.00247714 (* 0.0909091 = 0.000225195 loss)
I0607 01:42:37.134008 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.000899483 (* 0.0909091 = 8.17711e-05 loss)
I0607 01:42:37.134022 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.000404039 (* 0.0909091 = 3.67308e-05 loss)
I0607 01:42:37.134035 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 0.000241604 (* 0.0909091 = 2.1964e-05 loss)
I0607 01:42:37.134047 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.625
I0607 01:42:37.134059 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.625
I0607 01:42:37.134080 32403 solver.cpp:245] Train net output #149: total_confidence = 0.465867
I0607 01:42:37.134094 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.418985
I0607 01:42:37.134109 32403 sgd_solver.cpp:106] Iteration 3000, lr = 0.001
I0607 01:42:56.940446 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.8771 > 30) by scale factor 0.971594
I0607 01:43:10.913640 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.5605 > 30) by scale factor 0.868044
I0607 01:43:52.916779 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.3741 > 30) by scale factor 0.926667
I0607 01:44:49.685945 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.6778 > 30) by scale factor 0.918055
I0607 01:45:21.519675 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 50.2656 > 30) by scale factor 0.59683
I0607 01:45:40.166077 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.9444 > 30) by scale factor 0.858507
I0607 01:49:05.791643 32403 solver.cpp:229] Iteration 3500, loss = 4.35343
I0607 01:49:05.791785 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.518519
I0607 01:49:05.791806 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.625
I0607 01:49:05.791821 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.625
I0607 01:49:05.791832 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.5
I0607 01:49:05.791846 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.625
I0607 01:49:05.791857 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.625
I0607 01:49:05.791869 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.75
I0607 01:49:05.791884 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.5
I0607 01:49:05.791898 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.875
I0607 01:49:05.791909 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.875
I0607 01:49:05.791921 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.75
I0607 01:49:05.791934 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.75
I0607 01:49:05.791946 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 01:49:05.791959 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 1
I0607 01:49:05.791970 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 1
I0607 01:49:05.791982 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 01:49:05.791995 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 01:49:05.792006 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 01:49:05.792017 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 01:49:05.792029 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 01:49:05.792042 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 01:49:05.792053 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 01:49:05.792065 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 01:49:05.792076 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.835227
I0607 01:49:05.792089 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.777778
I0607 01:49:05.792106 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.57004 (* 0.3 = 0.471011 loss)
I0607 01:49:05.792120 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.554107 (* 0.3 = 0.166232 loss)
I0607 01:49:05.792135 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 1.06661 (* 0.0272727 = 0.0290894 loss)
I0607 01:49:05.792150 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.28579 (* 0.0272727 = 0.035067 loss)
I0607 01:49:05.792163 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.37786 (* 0.0272727 = 0.0375781 loss)
I0607 01:49:05.792177 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.06196 (* 0.0272727 = 0.0289624 loss)
I0607 01:49:05.792191 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.22479 (* 0.0272727 = 0.0334033 loss)
I0607 01:49:05.792204 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.41286 (* 0.0272727 = 0.0385326 loss)
I0607 01:49:05.792218 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 1.51088 (* 0.0272727 = 0.0412057 loss)
I0607 01:49:05.792232 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.630736 (* 0.0272727 = 0.0172019 loss)
I0607 01:49:05.792248 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.730574 (* 0.0272727 = 0.0199247 loss)
I0607 01:49:05.792261 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.627348 (* 0.0272727 = 0.0171095 loss)
I0607 01:49:05.792275 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.599985 (* 0.0272727 = 0.0163632 loss)
I0607 01:49:05.792289 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.562006 (* 0.0272727 = 0.0153274 loss)
I0607 01:49:05.792330 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.0944055 (* 0.0272727 = 0.00257469 loss)
I0607 01:49:05.792346 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.0328164 (* 0.0272727 = 0.000894992 loss)
I0607 01:49:05.792359 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.00964885 (* 0.0272727 = 0.000263151 loss)
I0607 01:49:05.792374 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.00199457 (* 0.0272727 = 5.43974e-05 loss)
I0607 01:49:05.792388 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.000385932 (* 0.0272727 = 1.05254e-05 loss)
I0607 01:49:05.792402 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.000200078 (* 0.0272727 = 5.45667e-06 loss)
I0607 01:49:05.792418 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 7.98111e-05 (* 0.0272727 = 2.17667e-06 loss)
I0607 01:49:05.792431 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 6.97221e-05 (* 0.0272727 = 1.90151e-06 loss)
I0607 01:49:05.792445 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.000131396 (* 0.0272727 = 3.58352e-06 loss)
I0607 01:49:05.792459 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 7.36829e-05 (* 0.0272727 = 2.00953e-06 loss)
I0607 01:49:05.792472 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.666667
I0607 01:49:05.792485 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.875
I0607 01:49:05.792497 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.875
I0607 01:49:05.792510 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.875
I0607 01:49:05.792521 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.5
I0607 01:49:05.792533 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.625
I0607 01:49:05.792546 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.875
I0607 01:49:05.792557 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.75
I0607 01:49:05.792569 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.875
I0607 01:49:05.792582 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.875
I0607 01:49:05.792594 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.75
I0607 01:49:05.792605 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.875
I0607 01:49:05.792618 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.875
I0607 01:49:05.792630 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 1
I0607 01:49:05.792641 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 1
I0607 01:49:05.792654 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 01:49:05.792665 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 01:49:05.792677 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 01:49:05.792690 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 01:49:05.792701 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 01:49:05.792712 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 01:49:05.792723 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 01:49:05.792735 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 01:49:05.792747 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.869318
I0607 01:49:05.792758 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.87037
I0607 01:49:05.792773 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.02016 (* 0.3 = 0.306048 loss)
I0607 01:49:05.792786 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.380084 (* 0.3 = 0.114025 loss)
I0607 01:49:05.792804 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.464718 (* 0.0272727 = 0.0126741 loss)
I0607 01:49:05.792819 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.538185 (* 0.0272727 = 0.0146778 loss)
I0607 01:49:05.792845 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.671251 (* 0.0272727 = 0.0183068 loss)
I0607 01:49:05.792856 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 0.94903 (* 0.0272727 = 0.0258826 loss)
I0607 01:49:05.792865 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.17332 (* 0.0272727 = 0.0319997 loss)
I0607 01:49:05.792881 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 0.864436 (* 0.0272727 = 0.0235755 loss)
I0607 01:49:05.792894 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.886144 (* 0.0272727 = 0.0241676 loss)
I0607 01:49:05.792908 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.54528 (* 0.0272727 = 0.0148713 loss)
I0607 01:49:05.792922 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.640281 (* 0.0272727 = 0.0174622 loss)
I0607 01:49:05.792938 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.805639 (* 0.0272727 = 0.021972 loss)
I0607 01:49:05.792953 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.442667 (* 0.0272727 = 0.0120727 loss)
I0607 01:49:05.792966 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.878527 (* 0.0272727 = 0.0239598 loss)
I0607 01:49:05.792981 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.150591 (* 0.0272727 = 0.00410704 loss)
I0607 01:49:05.792995 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.036747 (* 0.0272727 = 0.00100219 loss)
I0607 01:49:05.793009 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.00553642 (* 0.0272727 = 0.000150993 loss)
I0607 01:49:05.793023 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.0017609 (* 0.0272727 = 4.80246e-05 loss)
I0607 01:49:05.793037 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.000566087 (* 0.0272727 = 1.54387e-05 loss)
I0607 01:49:05.793051 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.000111666 (* 0.0272727 = 3.04544e-06 loss)
I0607 01:49:05.793066 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.000150553 (* 0.0272727 = 4.106e-06 loss)
I0607 01:49:05.793081 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 4.49686e-05 (* 0.0272727 = 1.22642e-06 loss)
I0607 01:49:05.793095 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 1.6645e-05 (* 0.0272727 = 4.53954e-07 loss)
I0607 01:49:05.793109 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 2.6824e-05 (* 0.0272727 = 7.31563e-07 loss)
I0607 01:49:05.793135 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.777778
I0607 01:49:05.793149 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 0.75
I0607 01:49:05.793161 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 1
I0607 01:49:05.793172 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.875
I0607 01:49:05.793185 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.75
I0607 01:49:05.793196 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.75
I0607 01:49:05.793208 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 1
I0607 01:49:05.793220 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 1
I0607 01:49:05.793231 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 1
I0607 01:49:05.793243 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.875
I0607 01:49:05.793256 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 1
I0607 01:49:05.793267 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.875
I0607 01:49:05.793278 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.875
I0607 01:49:05.793290 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 1
I0607 01:49:05.793301 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 01:49:05.793313 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 01:49:05.793335 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 01:49:05.793349 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 01:49:05.793360 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 01:49:05.793372 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 01:49:05.793385 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 01:49:05.793395 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 01:49:05.793407 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 01:49:05.793418 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.914773
I0607 01:49:05.793431 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 1
I0607 01:49:05.793444 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.636437 (* 1 = 0.636437 loss)
I0607 01:49:05.793458 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.246685 (* 1 = 0.246685 loss)
I0607 01:49:05.793473 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.316652 (* 0.0909091 = 0.0287865 loss)
I0607 01:49:05.793488 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.117002 (* 0.0909091 = 0.0106365 loss)
I0607 01:49:05.793501 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.474494 (* 0.0909091 = 0.0431358 loss)
I0607 01:49:05.793515 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.920005 (* 0.0909091 = 0.0836369 loss)
I0607 01:49:05.793529 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.758027 (* 0.0909091 = 0.0689116 loss)
I0607 01:49:05.793542 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.358244 (* 0.0909091 = 0.0325677 loss)
I0607 01:49:05.793556 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.250546 (* 0.0909091 = 0.0227769 loss)
I0607 01:49:05.793570 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.236611 (* 0.0909091 = 0.0215101 loss)
I0607 01:49:05.793584 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.403544 (* 0.0909091 = 0.0366858 loss)
I0607 01:49:05.793598 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.352596 (* 0.0909091 = 0.0320542 loss)
I0607 01:49:05.793612 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.34985 (* 0.0909091 = 0.0318046 loss)
I0607 01:49:05.793628 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.395642 (* 0.0909091 = 0.0359674 loss)
I0607 01:49:05.793644 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.0441816 (* 0.0909091 = 0.00401651 loss)
I0607 01:49:05.793658 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.0097261 (* 0.0909091 = 0.000884191 loss)
I0607 01:49:05.793671 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.00199459 (* 0.0909091 = 0.000181326 loss)
I0607 01:49:05.793685 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.00063427 (* 0.0909091 = 5.76609e-05 loss)
I0607 01:49:05.793699 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.000215807 (* 0.0909091 = 1.96188e-05 loss)
I0607 01:49:05.793714 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.000104655 (* 0.0909091 = 9.51405e-06 loss)
I0607 01:49:05.793727 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 7.03969e-05 (* 0.0909091 = 6.39972e-06 loss)
I0607 01:49:05.793741 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 7.63289e-05 (* 0.0909091 = 6.93899e-06 loss)
I0607 01:49:05.793756 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 6.17086e-05 (* 0.0909091 = 5.60987e-06 loss)
I0607 01:49:05.793769 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 6.27604e-05 (* 0.0909091 = 5.70549e-06 loss)
I0607 01:49:05.793781 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.375
I0607 01:49:05.793793 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.375
I0607 01:49:05.793814 32403 solver.cpp:245] Train net output #149: total_confidence = 0.347761
I0607 01:49:05.793828 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.356107
I0607 01:49:05.793840 32403 sgd_solver.cpp:106] Iteration 3500, lr = 0.001
I0607 01:51:28.088018 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.3601 > 30) by scale factor 0.95663
I0607 01:51:59.099689 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 49.3691 > 30) by scale factor 0.607668
I0607 01:55:33.735272 32403 solver.cpp:229] Iteration 4000, loss = 4.16428
I0607 01:55:33.735458 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.6
I0607 01:55:33.735481 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.75
I0607 01:55:33.735493 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.875
I0607 01:55:33.735507 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.625
I0607 01:55:33.735518 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.5
I0607 01:55:33.735530 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.75
I0607 01:55:33.735543 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.625
I0607 01:55:33.735554 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.75
I0607 01:55:33.735568 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.875
I0607 01:55:33.735579 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 1
I0607 01:55:33.735591 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.875
I0607 01:55:33.735605 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.875
I0607 01:55:33.735616 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 01:55:33.735630 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 1
I0607 01:55:33.735641 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 1
I0607 01:55:33.735652 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 01:55:33.735664 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 01:55:33.735677 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 01:55:33.735688 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 01:55:33.735702 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 01:55:33.735713 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 01:55:33.735724 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 01:55:33.735736 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 01:55:33.735749 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.886364
I0607 01:55:33.735761 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.8
I0607 01:55:33.735777 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.38727 (* 0.3 = 0.41618 loss)
I0607 01:55:33.735792 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.421738 (* 0.3 = 0.126522 loss)
I0607 01:55:33.735807 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.823164 (* 0.0272727 = 0.0224499 loss)
I0607 01:55:33.735821 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 0.701339 (* 0.0272727 = 0.0191274 loss)
I0607 01:55:33.735836 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.62878 (* 0.0272727 = 0.0444213 loss)
I0607 01:55:33.735851 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.60252 (* 0.0272727 = 0.0437051 loss)
I0607 01:55:33.735864 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.18089 (* 0.0272727 = 0.0322061 loss)
I0607 01:55:33.735882 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 0.852906 (* 0.0272727 = 0.0232611 loss)
I0607 01:55:33.735895 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 0.622128 (* 0.0272727 = 0.0169671 loss)
I0607 01:55:33.735910 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.354401 (* 0.0272727 = 0.00966548 loss)
I0607 01:55:33.735925 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.258633 (* 0.0272727 = 0.00705362 loss)
I0607 01:55:33.735939 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.444786 (* 0.0272727 = 0.0121305 loss)
I0607 01:55:33.735954 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.49225 (* 0.0272727 = 0.013425 loss)
I0607 01:55:33.735968 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.368551 (* 0.0272727 = 0.0100514 loss)
I0607 01:55:33.735982 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.0717179 (* 0.0272727 = 0.00195594 loss)
I0607 01:55:33.736028 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.0526367 (* 0.0272727 = 0.00143555 loss)
I0607 01:55:33.736043 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.0477881 (* 0.0272727 = 0.00130331 loss)
I0607 01:55:33.736058 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.0241529 (* 0.0272727 = 0.000658716 loss)
I0607 01:55:33.736073 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.00810096 (* 0.0272727 = 0.000220935 loss)
I0607 01:55:33.736086 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.00734096 (* 0.0272727 = 0.000200208 loss)
I0607 01:55:33.736100 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.0041991 (* 0.0272727 = 0.000114521 loss)
I0607 01:55:33.736115 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.00163098 (* 0.0272727 = 4.44813e-05 loss)
I0607 01:55:33.736129 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.000834946 (* 0.0272727 = 2.27712e-05 loss)
I0607 01:55:33.736145 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.00131517 (* 0.0272727 = 3.58683e-05 loss)
I0607 01:55:33.736157 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.666667
I0607 01:55:33.736171 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.75
I0607 01:55:33.736182 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.75
I0607 01:55:33.736194 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.75
I0607 01:55:33.736207 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.5
I0607 01:55:33.736218 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.75
I0607 01:55:33.736230 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.625
I0607 01:55:33.736243 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.875
I0607 01:55:33.736254 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.875
I0607 01:55:33.736266 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.875
I0607 01:55:33.736279 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.875
I0607 01:55:33.736290 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.875
I0607 01:55:33.736302 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.875
I0607 01:55:33.736315 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 1
I0607 01:55:33.736326 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 1
I0607 01:55:33.736338 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 01:55:33.736349 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 01:55:33.736361 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 01:55:33.736373 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 01:55:33.736385 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 01:55:33.736397 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 01:55:33.736408 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 01:55:33.736429 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 01:55:33.736441 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.914773
I0607 01:55:33.736456 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.822222
I0607 01:55:33.736471 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.0731 (* 0.3 = 0.32193 loss)
I0607 01:55:33.736495 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.297707 (* 0.3 = 0.0893122 loss)
I0607 01:55:33.736508 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.721425 (* 0.0272727 = 0.0196752 loss)
I0607 01:55:33.736523 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.704277 (* 0.0272727 = 0.0192076 loss)
I0607 01:55:33.736548 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.775787 (* 0.0272727 = 0.0211578 loss)
I0607 01:55:33.736563 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.28522 (* 0.0272727 = 0.0350514 loss)
I0607 01:55:33.736577 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 0.689767 (* 0.0272727 = 0.0188118 loss)
I0607 01:55:33.736591 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 0.61361 (* 0.0272727 = 0.0167348 loss)
I0607 01:55:33.736605 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.344544 (* 0.0272727 = 0.00939665 loss)
I0607 01:55:33.736620 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.409641 (* 0.0272727 = 0.011172 loss)
I0607 01:55:33.736634 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.353259 (* 0.0272727 = 0.00963433 loss)
I0607 01:55:33.736649 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.56615 (* 0.0272727 = 0.0154404 loss)
I0607 01:55:33.736663 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.473918 (* 0.0272727 = 0.012925 loss)
I0607 01:55:33.736677 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.616229 (* 0.0272727 = 0.0168062 loss)
I0607 01:55:33.736691 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.00237714 (* 0.0272727 = 6.48311e-05 loss)
I0607 01:55:33.736706 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.00142908 (* 0.0272727 = 3.89749e-05 loss)
I0607 01:55:33.736721 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.000771049 (* 0.0272727 = 2.10286e-05 loss)
I0607 01:55:33.736734 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.00111134 (* 0.0272727 = 3.03093e-05 loss)
I0607 01:55:33.736745 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.000985809 (* 0.0272727 = 2.68857e-05 loss)
I0607 01:55:33.736755 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.00546858 (* 0.0272727 = 0.000149143 loss)
I0607 01:55:33.736765 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.00110653 (* 0.0272727 = 3.01781e-05 loss)
I0607 01:55:33.736780 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.000975655 (* 0.0272727 = 2.66088e-05 loss)
I0607 01:55:33.736794 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.00106281 (* 0.0272727 = 2.89857e-05 loss)
I0607 01:55:33.736809 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 0.000388508 (* 0.0272727 = 1.05957e-05 loss)
I0607 01:55:33.736821 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.777778
I0607 01:55:33.736834 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 0.875
I0607 01:55:33.736845 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.875
I0607 01:55:33.736857 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.625
I0607 01:55:33.736870 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.875
I0607 01:55:33.736881 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 1
I0607 01:55:33.736893 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.875
I0607 01:55:33.736906 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 1
I0607 01:55:33.736917 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.875
I0607 01:55:33.736932 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.875
I0607 01:55:33.736943 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.875
I0607 01:55:33.736955 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.875
I0607 01:55:33.736968 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.875
I0607 01:55:33.736985 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 1
I0607 01:55:33.736997 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 01:55:33.737009 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 01:55:33.737030 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 01:55:33.737048 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 01:55:33.737061 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 01:55:33.737071 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 01:55:33.737083 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 01:55:33.737095 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 01:55:33.737107 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 01:55:33.737130 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.943182
I0607 01:55:33.737146 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.933333
I0607 01:55:33.737160 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.701819 (* 1 = 0.701819 loss)
I0607 01:55:33.737174 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.199869 (* 1 = 0.199869 loss)
I0607 01:55:33.737188 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.481698 (* 0.0909091 = 0.0437908 loss)
I0607 01:55:33.737203 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.310242 (* 0.0909091 = 0.0282039 loss)
I0607 01:55:33.737217 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.991739 (* 0.0909091 = 0.0901581 loss)
I0607 01:55:33.737231 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.507901 (* 0.0909091 = 0.0461728 loss)
I0607 01:55:33.737246 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.177654 (* 0.0909091 = 0.0161504 loss)
I0607 01:55:33.737259 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.924342 (* 0.0909091 = 0.0840311 loss)
I0607 01:55:33.737273 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.019325 (* 0.0909091 = 0.00175682 loss)
I0607 01:55:33.737288 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.536353 (* 0.0909091 = 0.0487594 loss)
I0607 01:55:33.737301 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.21144 (* 0.0909091 = 0.0192218 loss)
I0607 01:55:33.737315 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.146373 (* 0.0909091 = 0.0133066 loss)
I0607 01:55:33.737329 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.199887 (* 0.0909091 = 0.0181715 loss)
I0607 01:55:33.737344 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.274527 (* 0.0909091 = 0.024957 loss)
I0607 01:55:33.737359 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.0212143 (* 0.0909091 = 0.00192857 loss)
I0607 01:55:33.737372 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.00777487 (* 0.0909091 = 0.000706806 loss)
I0607 01:55:33.737387 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.00172071 (* 0.0909091 = 0.000156428 loss)
I0607 01:55:33.737401 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.000330198 (* 0.0909091 = 3.0018e-05 loss)
I0607 01:55:33.737416 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.000187455 (* 0.0909091 = 1.70414e-05 loss)
I0607 01:55:33.737431 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.000180342 (* 0.0909091 = 1.63947e-05 loss)
I0607 01:55:33.737444 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.000167022 (* 0.0909091 = 1.51838e-05 loss)
I0607 01:55:33.737459 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.000157348 (* 0.0909091 = 1.43044e-05 loss)
I0607 01:55:33.737473 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.00021753 (* 0.0909091 = 1.97755e-05 loss)
I0607 01:55:33.737488 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 0.000204776 (* 0.0909091 = 1.8616e-05 loss)
I0607 01:55:33.737500 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.625
I0607 01:55:33.737515 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.625
I0607 01:55:33.737534 32403 solver.cpp:245] Train net output #149: total_confidence = 0.52339
I0607 01:55:33.737550 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.522814
I0607 01:55:33.737563 32403 sgd_solver.cpp:106] Iteration 4000, lr = 0.001
I0607 01:56:19.876839 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.934 > 30) by scale factor 0.939437
I0607 01:56:44.692603 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 61.1428 > 30) by scale factor 0.490655
I0607 01:57:28.146356 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.3086 > 30) by scale factor 0.958204
I0607 01:57:46.753665 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 40.8963 > 30) by scale factor 0.733563
I0607 01:58:02.236335 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.7299 > 30) by scale factor 0.795126
I0607 02:00:03.248699 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 36.2607 > 30) by scale factor 0.827342
I0607 02:00:13.315472 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.8262 > 30) by scale factor 0.86142
I0607 02:02:01.479104 32403 solver.cpp:229] Iteration 4500, loss = 4.18739
I0607 02:02:01.479240 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.657895
I0607 02:02:01.479261 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.875
I0607 02:02:01.479274 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.625
I0607 02:02:01.479287 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.375
I0607 02:02:01.479300 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.625
I0607 02:02:01.479311 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.5
I0607 02:02:01.479326 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.625
I0607 02:02:01.479337 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.75
I0607 02:02:01.479351 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 1
I0607 02:02:01.479362 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 1
I0607 02:02:01.479374 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 1
I0607 02:02:01.479387 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 1
I0607 02:02:01.479398 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 1
I0607 02:02:01.479409 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 1
I0607 02:02:01.479421 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 1
I0607 02:02:01.479434 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 02:02:01.479445 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 02:02:01.479457 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 02:02:01.479470 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 02:02:01.479480 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 02:02:01.479492 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 02:02:01.479506 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 02:02:01.479518 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 02:02:01.479531 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.909091
I0607 02:02:01.479542 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.789474
I0607 02:02:01.479558 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.19984 (* 0.3 = 0.359952 loss)
I0607 02:02:01.479573 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.311401 (* 0.3 = 0.0934202 loss)
I0607 02:02:01.479588 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.513075 (* 0.0272727 = 0.0139929 loss)
I0607 02:02:01.479603 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.42226 (* 0.0272727 = 0.0387888 loss)
I0607 02:02:01.479616 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.76571 (* 0.0272727 = 0.0481556 loss)
I0607 02:02:01.479630 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.54718 (* 0.0272727 = 0.0421959 loss)
I0607 02:02:01.479645 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.88283 (* 0.0272727 = 0.05135 loss)
I0607 02:02:01.479658 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 0.535917 (* 0.0272727 = 0.0146159 loss)
I0607 02:02:01.479673 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 0.293051 (* 0.0272727 = 0.00799229 loss)
I0607 02:02:01.479687 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.0531653 (* 0.0272727 = 0.00144996 loss)
I0607 02:02:01.479702 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.00444089 (* 0.0272727 = 0.000121115 loss)
I0607 02:02:01.479717 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.00134022 (* 0.0272727 = 3.65513e-05 loss)
I0607 02:02:01.479732 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.000880313 (* 0.0272727 = 2.40085e-05 loss)
I0607 02:02:01.479745 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.00125922 (* 0.0272727 = 3.43425e-05 loss)
I0607 02:02:01.479789 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.000939586 (* 0.0272727 = 2.56251e-05 loss)
I0607 02:02:01.479805 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.000942743 (* 0.0272727 = 2.57112e-05 loss)
I0607 02:02:01.479820 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.00104036 (* 0.0272727 = 2.83733e-05 loss)
I0607 02:02:01.479833 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.000359643 (* 0.0272727 = 9.80846e-06 loss)
I0607 02:02:01.479847 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.00155138 (* 0.0272727 = 4.23103e-05 loss)
I0607 02:02:01.479861 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.000718986 (* 0.0272727 = 1.96087e-05 loss)
I0607 02:02:01.479878 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.000783752 (* 0.0272727 = 2.1375e-05 loss)
I0607 02:02:01.479893 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.000869414 (* 0.0272727 = 2.37113e-05 loss)
I0607 02:02:01.479907 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.00116992 (* 0.0272727 = 3.19069e-05 loss)
I0607 02:02:01.479921 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.0021744 (* 0.0272727 = 5.93019e-05 loss)
I0607 02:02:01.479934 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.736842
I0607 02:02:01.479948 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.875
I0607 02:02:01.479960 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.625
I0607 02:02:01.479972 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.75
I0607 02:02:01.479984 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.875
I0607 02:02:01.479997 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.875
I0607 02:02:01.480010 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.75
I0607 02:02:01.480021 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.875
I0607 02:02:01.480033 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 1
I0607 02:02:01.480046 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 1
I0607 02:02:01.480057 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 1
I0607 02:02:01.480068 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 1
I0607 02:02:01.480080 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 1
I0607 02:02:01.480092 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 1
I0607 02:02:01.480103 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 1
I0607 02:02:01.480114 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 02:02:01.480126 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 02:02:01.480139 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 02:02:01.480150 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 02:02:01.480162 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 02:02:01.480173 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 02:02:01.480185 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 02:02:01.480197 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 02:02:01.480208 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.9375
I0607 02:02:01.480221 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.921053
I0607 02:02:01.480235 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 0.913001 (* 0.3 = 0.2739 loss)
I0607 02:02:01.480252 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.237742 (* 0.3 = 0.0713225 loss)
I0607 02:02:01.480267 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.531477 (* 0.0272727 = 0.0144948 loss)
I0607 02:02:01.480280 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 1.72021 (* 0.0272727 = 0.0469149 loss)
I0607 02:02:01.480306 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.699888 (* 0.0272727 = 0.0190879 loss)
I0607 02:02:01.480321 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 0.676508 (* 0.0272727 = 0.0184502 loss)
I0607 02:02:01.480335 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 0.528718 (* 0.0272727 = 0.0144196 loss)
I0607 02:02:01.480350 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 0.540004 (* 0.0272727 = 0.0147274 loss)
I0607 02:02:01.480365 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.298006 (* 0.0272727 = 0.00812744 loss)
I0607 02:02:01.480379 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.0534779 (* 0.0272727 = 0.00145849 loss)
I0607 02:02:01.480393 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.00330581 (* 0.0272727 = 9.01583e-05 loss)
I0607 02:02:01.480407 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.00031272 (* 0.0272727 = 8.52872e-06 loss)
I0607 02:02:01.480422 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 8.91149e-05 (* 0.0272727 = 2.43041e-06 loss)
I0607 02:02:01.480437 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 7.58486e-05 (* 0.0272727 = 2.0686e-06 loss)
I0607 02:02:01.480450 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.000143482 (* 0.0272727 = 3.91315e-06 loss)
I0607 02:02:01.480463 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 7.40855e-05 (* 0.0272727 = 2.02051e-06 loss)
I0607 02:02:01.480479 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 7.1179e-05 (* 0.0272727 = 1.94125e-06 loss)
I0607 02:02:01.480492 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.000150269 (* 0.0272727 = 4.09823e-06 loss)
I0607 02:02:01.480506 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.000135203 (* 0.0272727 = 3.68736e-06 loss)
I0607 02:02:01.480521 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.00013228 (* 0.0272727 = 3.60763e-06 loss)
I0607 02:02:01.480535 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.000169488 (* 0.0272727 = 4.6224e-06 loss)
I0607 02:02:01.480550 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.000213678 (* 0.0272727 = 5.82757e-06 loss)
I0607 02:02:01.480563 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.000205976 (* 0.0272727 = 5.61754e-06 loss)
I0607 02:02:01.480577 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 0.000137986 (* 0.0272727 = 3.76324e-06 loss)
I0607 02:02:01.480589 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.789474
I0607 02:02:01.480602 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 0.75
I0607 02:02:01.480614 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.625
I0607 02:02:01.480626 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.875
I0607 02:02:01.480638 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.875
I0607 02:02:01.480650 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.875
I0607 02:02:01.480662 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.875
I0607 02:02:01.480674 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.875
I0607 02:02:01.480686 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 1
I0607 02:02:01.480695 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 1
I0607 02:02:01.480703 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 1
I0607 02:02:01.480710 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 1
I0607 02:02:01.480726 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 1
I0607 02:02:01.480737 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 1
I0607 02:02:01.480749 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 02:02:01.480762 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 02:02:01.480782 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 02:02:01.480797 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 02:02:01.480808 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 02:02:01.480819 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 02:02:01.480831 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 02:02:01.480844 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 02:02:01.480854 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 02:02:01.480866 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.948864
I0607 02:02:01.480878 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.868421
I0607 02:02:01.480893 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.671727 (* 1 = 0.671727 loss)
I0607 02:02:01.480907 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.168736 (* 1 = 0.168736 loss)
I0607 02:02:01.480921 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.697933 (* 0.0909091 = 0.0634484 loss)
I0607 02:02:01.480937 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.897975 (* 0.0909091 = 0.0816341 loss)
I0607 02:02:01.480952 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.414066 (* 0.0909091 = 0.0376424 loss)
I0607 02:02:01.480965 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.595047 (* 0.0909091 = 0.0540952 loss)
I0607 02:02:01.480979 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.381212 (* 0.0909091 = 0.0346556 loss)
I0607 02:02:01.480993 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.320984 (* 0.0909091 = 0.0291803 loss)
I0607 02:02:01.481008 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.307326 (* 0.0909091 = 0.0279387 loss)
I0607 02:02:01.481022 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.0425477 (* 0.0909091 = 0.00386798 loss)
I0607 02:02:01.481035 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.00576733 (* 0.0909091 = 0.000524303 loss)
I0607 02:02:01.481050 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.00117032 (* 0.0909091 = 0.000106393 loss)
I0607 02:02:01.481063 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.000640813 (* 0.0909091 = 5.82557e-05 loss)
I0607 02:02:01.481077 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.000160399 (* 0.0909091 = 1.45817e-05 loss)
I0607 02:02:01.481091 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.000113058 (* 0.0909091 = 1.0278e-05 loss)
I0607 02:02:01.481106 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 7.1705e-05 (* 0.0909091 = 6.51864e-06 loss)
I0607 02:02:01.481132 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 4.83313e-05 (* 0.0909091 = 4.39375e-06 loss)
I0607 02:02:01.481148 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 2.73304e-05 (* 0.0909091 = 2.48459e-06 loss)
I0607 02:02:01.481163 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 4.30459e-05 (* 0.0909091 = 3.91326e-06 loss)
I0607 02:02:01.481176 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 4.42223e-05 (* 0.0909091 = 4.02021e-06 loss)
I0607 02:02:01.481190 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 5.12271e-05 (* 0.0909091 = 4.65701e-06 loss)
I0607 02:02:01.481204 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 4.54209e-05 (* 0.0909091 = 4.12918e-06 loss)
I0607 02:02:01.481218 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 3.86705e-05 (* 0.0909091 = 3.5155e-06 loss)
I0607 02:02:01.481232 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 4.04885e-05 (* 0.0909091 = 3.68078e-06 loss)
I0607 02:02:01.481245 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.625
I0607 02:02:01.481256 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.5
I0607 02:02:01.481279 32403 solver.cpp:245] Train net output #149: total_confidence = 0.580609
I0607 02:02:01.481293 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.498801
I0607 02:02:01.481309 32403 sgd_solver.cpp:106] Iteration 4500, lr = 0.001
I0607 02:03:42.723737 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.6197 > 30) by scale factor 0.797455
I0607 02:05:30.468489 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 43.9878 > 30) by scale factor 0.682007
I0607 02:06:13.106956 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.5303 > 30) by scale factor 0.982629
I0607 02:06:20.851640 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.2472 > 30) by scale factor 0.80543
I0607 02:06:48.763847 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.3105 > 30) by scale factor 0.874367
I0607 02:07:24.478458 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.6871 > 30) by scale factor 0.977609
I0607 02:08:28.918543 32403 solver.cpp:338] Iteration 5000, Testing net (#0)
I0607 02:09:27.142835 32403 solver.cpp:393] Test loss: 2.61234
I0607 02:09:27.142969 32403 solver.cpp:406] Test net output #0: loss1/accuracy = 0.622352
I0607 02:09:27.142998 32403 solver.cpp:406] Test net output #1: loss1/accuracy01 = 0.786
I0607 02:09:27.143020 32403 solver.cpp:406] Test net output #2: loss1/accuracy02 = 0.653
I0607 02:09:27.143043 32403 solver.cpp:406] Test net output #3: loss1/accuracy03 = 0.535
I0607 02:09:27.143064 32403 solver.cpp:406] Test net output #4: loss1/accuracy04 = 0.473
I0607 02:09:27.143085 32403 solver.cpp:406] Test net output #5: loss1/accuracy05 = 0.531
I0607 02:09:27.143103 32403 solver.cpp:406] Test net output #6: loss1/accuracy06 = 0.717
I0607 02:09:27.143123 32403 solver.cpp:406] Test net output #7: loss1/accuracy07 = 0.853
I0607 02:09:27.143153 32403 solver.cpp:406] Test net output #8: loss1/accuracy08 = 0.92
I0607 02:09:27.143185 32403 solver.cpp:406] Test net output #9: loss1/accuracy09 = 0.969
I0607 02:09:27.143206 32403 solver.cpp:406] Test net output #10: loss1/accuracy10 = 0.988
I0607 02:09:27.143225 32403 solver.cpp:406] Test net output #11: loss1/accuracy11 = 0.997
I0607 02:09:27.143245 32403 solver.cpp:406] Test net output #12: loss1/accuracy12 = 0.999
I0607 02:09:27.143263 32403 solver.cpp:406] Test net output #13: loss1/accuracy13 = 1
I0607 02:09:27.143282 32403 solver.cpp:406] Test net output #14: loss1/accuracy14 = 1
I0607 02:09:27.143301 32403 solver.cpp:406] Test net output #15: loss1/accuracy15 = 1
I0607 02:09:27.143321 32403 solver.cpp:406] Test net output #16: loss1/accuracy16 = 1
I0607 02:09:27.143338 32403 solver.cpp:406] Test net output #17: loss1/accuracy17 = 1
I0607 02:09:27.143357 32403 solver.cpp:406] Test net output #18: loss1/accuracy18 = 1
I0607 02:09:27.143378 32403 solver.cpp:406] Test net output #19: loss1/accuracy19 = 1
I0607 02:09:27.143395 32403 solver.cpp:406] Test net output #20: loss1/accuracy20 = 1
I0607 02:09:27.143414 32403 solver.cpp:406] Test net output #21: loss1/accuracy21 = 1
I0607 02:09:27.143436 32403 solver.cpp:406] Test net output #22: loss1/accuracy22 = 1
I0607 02:09:27.143455 32403 solver.cpp:406] Test net output #23: loss1/accuracy_incl_empty = 0.891002
I0607 02:09:27.143476 32403 solver.cpp:406] Test net output #24: loss1/accuracy_top3 = 0.843025
I0607 02:09:27.143501 32403 solver.cpp:406] Test net output #25: loss1/cross_entropy_loss = 1.36898 (* 0.3 = 0.410694 loss)
I0607 02:09:27.143525 32403 solver.cpp:406] Test net output #26: loss1/cross_entropy_loss_incl_empty = 0.398625 (* 0.3 = 0.119588 loss)
I0607 02:09:27.143549 32403 solver.cpp:406] Test net output #27: loss1/loss01 = 0.917259 (* 0.0272727 = 0.0250162 loss)
I0607 02:09:27.143573 32403 solver.cpp:406] Test net output #28: loss1/loss02 = 1.35055 (* 0.0272727 = 0.0368332 loss)
I0607 02:09:27.143595 32403 solver.cpp:406] Test net output #29: loss1/loss03 = 1.69251 (* 0.0272727 = 0.0461592 loss)
I0607 02:09:27.143618 32403 solver.cpp:406] Test net output #30: loss1/loss04 = 1.83189 (* 0.0272727 = 0.0499606 loss)
I0607 02:09:27.143640 32403 solver.cpp:406] Test net output #31: loss1/loss05 = 1.59056 (* 0.0272727 = 0.0433788 loss)
I0607 02:09:27.143663 32403 solver.cpp:406] Test net output #32: loss1/loss06 = 1.00246 (* 0.0272727 = 0.0273399 loss)
I0607 02:09:27.143685 32403 solver.cpp:406] Test net output #33: loss1/loss07 = 0.525652 (* 0.0272727 = 0.014336 loss)
I0607 02:09:27.143709 32403 solver.cpp:406] Test net output #34: loss1/loss08 = 0.29335 (* 0.0272727 = 0.00800045 loss)
I0607 02:09:27.143733 32403 solver.cpp:406] Test net output #35: loss1/loss09 = 0.147346 (* 0.0272727 = 0.00401853 loss)
I0607 02:09:27.143755 32403 solver.cpp:406] Test net output #36: loss1/loss10 = 0.0708745 (* 0.0272727 = 0.00193294 loss)
I0607 02:09:27.143779 32403 solver.cpp:406] Test net output #37: loss1/loss11 = 0.0233996 (* 0.0272727 = 0.00063817 loss)
I0607 02:09:27.143802 32403 solver.cpp:406] Test net output #38: loss1/loss12 = 0.0140116 (* 0.0272727 = 0.000382134 loss)
I0607 02:09:27.143824 32403 solver.cpp:406] Test net output #39: loss1/loss13 = 0.00945804 (* 0.0272727 = 0.000257947 loss)
I0607 02:09:27.143882 32403 solver.cpp:406] Test net output #40: loss1/loss14 = 0.00737565 (* 0.0272727 = 0.000201154 loss)
I0607 02:09:27.143908 32403 solver.cpp:406] Test net output #41: loss1/loss15 = 0.00552067 (* 0.0272727 = 0.000150564 loss)
I0607 02:09:27.143934 32403 solver.cpp:406] Test net output #42: loss1/loss16 = 0.00446162 (* 0.0272727 = 0.00012168 loss)
I0607 02:09:27.143960 32403 solver.cpp:406] Test net output #43: loss1/loss17 = 0.00434735 (* 0.0272727 = 0.000118564 loss)
I0607 02:09:27.143985 32403 solver.cpp:406] Test net output #44: loss1/loss18 = 0.00402947 (* 0.0272727 = 0.000109895 loss)
I0607 02:09:27.144008 32403 solver.cpp:406] Test net output #45: loss1/loss19 = 0.00338127 (* 0.0272727 = 9.22166e-05 loss)
I0607 02:09:27.144033 32403 solver.cpp:406] Test net output #46: loss1/loss20 = 0.00336473 (* 0.0272727 = 9.17654e-05 loss)
I0607 02:09:27.144057 32403 solver.cpp:406] Test net output #47: loss1/loss21 = 0.00310117 (* 0.0272727 = 8.45773e-05 loss)
I0607 02:09:27.144083 32403 solver.cpp:406] Test net output #48: loss1/loss22 = 0.00276168 (* 0.0272727 = 7.53186e-05 loss)
I0607 02:09:27.144103 32403 solver.cpp:406] Test net output #49: loss2/accuracy = 0.765888
I0607 02:09:27.144122 32403 solver.cpp:406] Test net output #50: loss2/accuracy01 = 0.874
I0607 02:09:27.144142 32403 solver.cpp:406] Test net output #51: loss2/accuracy02 = 0.829
I0607 02:09:27.144161 32403 solver.cpp:406] Test net output #52: loss2/accuracy03 = 0.759
I0607 02:09:27.144181 32403 solver.cpp:406] Test net output #53: loss2/accuracy04 = 0.652
I0607 02:09:27.144206 32403 solver.cpp:406] Test net output #54: loss2/accuracy05 = 0.665
I0607 02:09:27.144227 32403 solver.cpp:406] Test net output #55: loss2/accuracy06 = 0.78
I0607 02:09:27.144245 32403 solver.cpp:406] Test net output #56: loss2/accuracy07 = 0.865
I0607 02:09:27.144264 32403 solver.cpp:406] Test net output #57: loss2/accuracy08 = 0.929
I0607 02:09:27.144284 32403 solver.cpp:406] Test net output #58: loss2/accuracy09 = 0.967
I0607 02:09:27.144302 32403 solver.cpp:406] Test net output #59: loss2/accuracy10 = 0.984
I0607 02:09:27.144322 32403 solver.cpp:406] Test net output #60: loss2/accuracy11 = 0.995
I0607 02:09:27.144341 32403 solver.cpp:406] Test net output #61: loss2/accuracy12 = 0.999
I0607 02:09:27.144361 32403 solver.cpp:406] Test net output #62: loss2/accuracy13 = 1
I0607 02:09:27.144378 32403 solver.cpp:406] Test net output #63: loss2/accuracy14 = 1
I0607 02:09:27.144398 32403 solver.cpp:406] Test net output #64: loss2/accuracy15 = 1
I0607 02:09:27.144417 32403 solver.cpp:406] Test net output #65: loss2/accuracy16 = 1
I0607 02:09:27.144435 32403 solver.cpp:406] Test net output #66: loss2/accuracy17 = 1
I0607 02:09:27.144454 32403 solver.cpp:406] Test net output #67: loss2/accuracy18 = 1
I0607 02:09:27.144474 32403 solver.cpp:406] Test net output #68: loss2/accuracy19 = 1
I0607 02:09:27.144491 32403 solver.cpp:406] Test net output #69: loss2/accuracy20 = 1
I0607 02:09:27.144510 32403 solver.cpp:406] Test net output #70: loss2/accuracy21 = 1
I0607 02:09:27.144528 32403 solver.cpp:406] Test net output #71: loss2/accuracy22 = 1
I0607 02:09:27.144546 32403 solver.cpp:406] Test net output #72: loss2/accuracy_incl_empty = 0.930637
I0607 02:09:27.144565 32403 solver.cpp:406] Test net output #73: loss2/accuracy_top3 = 0.914491
I0607 02:09:27.144588 32403 solver.cpp:406] Test net output #74: loss2/cross_entropy_loss = 0.925448 (* 0.3 = 0.277635 loss)
I0607 02:09:27.144611 32403 solver.cpp:406] Test net output #75: loss2/cross_entropy_loss_incl_empty = 0.267685 (* 0.3 = 0.0803055 loss)
I0607 02:09:27.144634 32403 solver.cpp:406] Test net output #76: loss2/loss01 = 0.5965 (* 0.0272727 = 0.0162682 loss)
I0607 02:09:27.144657 32403 solver.cpp:406] Test net output #77: loss2/loss02 = 0.819483 (* 0.0272727 = 0.0223495 loss)
I0607 02:09:27.144700 32403 solver.cpp:406] Test net output #78: loss2/loss03 = 0.997874 (* 0.0272727 = 0.0272147 loss)
I0607 02:09:27.144726 32403 solver.cpp:406] Test net output #79: loss2/loss04 = 1.22552 (* 0.0272727 = 0.0334233 loss)
I0607 02:09:27.144748 32403 solver.cpp:406] Test net output #80: loss2/loss05 = 1.11604 (* 0.0272727 = 0.0304376 loss)
I0607 02:09:27.144772 32403 solver.cpp:406] Test net output #81: loss2/loss06 = 0.751485 (* 0.0272727 = 0.020495 loss)
I0607 02:09:27.144795 32403 solver.cpp:406] Test net output #82: loss2/loss07 = 0.444735 (* 0.0272727 = 0.0121291 loss)
I0607 02:09:27.144819 32403 solver.cpp:406] Test net output #83: loss2/loss08 = 0.224137 (* 0.0272727 = 0.00611282 loss)
I0607 02:09:27.144842 32403 solver.cpp:406] Test net output #84: loss2/loss09 = 0.13303 (* 0.0272727 = 0.0036281 loss)
I0607 02:09:27.144866 32403 solver.cpp:406] Test net output #85: loss2/loss10 = 0.0642206 (* 0.0272727 = 0.00175147 loss)
I0607 02:09:27.144889 32403 solver.cpp:406] Test net output #86: loss2/loss11 = 0.0163907 (* 0.0272727 = 0.000447018 loss)
I0607 02:09:27.144913 32403 solver.cpp:406] Test net output #87: loss2/loss12 = 0.00666963 (* 0.0272727 = 0.000181899 loss)
I0607 02:09:27.144942 32403 solver.cpp:406] Test net output #88: loss2/loss13 = 0.00375302 (* 0.0272727 = 0.000102355 loss)
I0607 02:09:27.144965 32403 solver.cpp:406] Test net output #89: loss2/loss14 = 0.00228675 (* 0.0272727 = 6.23659e-05 loss)
I0607 02:09:27.144989 32403 solver.cpp:406] Test net output #90: loss2/loss15 = 0.00157399 (* 0.0272727 = 4.29269e-05 loss)
I0607 02:09:27.145015 32403 solver.cpp:406] Test net output #91: loss2/loss16 = 0.00126608 (* 0.0272727 = 3.45295e-05 loss)
I0607 02:09:27.145038 32403 solver.cpp:406] Test net output #92: loss2/loss17 = 0.00103579 (* 0.0272727 = 2.82489e-05 loss)
I0607 02:09:27.145063 32403 solver.cpp:406] Test net output #93: loss2/loss18 = 0.00111542 (* 0.0272727 = 3.04205e-05 loss)
I0607 02:09:27.145088 32403 solver.cpp:406] Test net output #94: loss2/loss19 = 0.000770504 (* 0.0272727 = 2.10137e-05 loss)
I0607 02:09:27.145112 32403 solver.cpp:406] Test net output #95: loss2/loss20 = 0.0009043 (* 0.0272727 = 2.46627e-05 loss)
I0607 02:09:27.145164 32403 solver.cpp:406] Test net output #96: loss2/loss21 = 0.000868332 (* 0.0272727 = 2.36818e-05 loss)
I0607 02:09:27.145190 32403 solver.cpp:406] Test net output #97: loss2/loss22 = 0.000827592 (* 0.0272727 = 2.25707e-05 loss)
I0607 02:09:27.145210 32403 solver.cpp:406] Test net output #98: loss3/accuracy = 0.847509
I0607 02:09:27.145231 32403 solver.cpp:406] Test net output #99: loss3/accuracy01 = 0.887
I0607 02:09:27.145254 32403 solver.cpp:406] Test net output #100: loss3/accuracy02 = 0.859
I0607 02:09:27.145275 32403 solver.cpp:406] Test net output #101: loss3/accuracy03 = 0.854
I0607 02:09:27.145295 32403 solver.cpp:406] Test net output #102: loss3/accuracy04 = 0.845
I0607 02:09:27.145315 32403 solver.cpp:406] Test net output #103: loss3/accuracy05 = 0.829
I0607 02:09:27.145334 32403 solver.cpp:406] Test net output #104: loss3/accuracy06 = 0.861
I0607 02:09:27.145354 32403 solver.cpp:406] Test net output #105: loss3/accuracy07 = 0.897
I0607 02:09:27.145375 32403 solver.cpp:406] Test net output #106: loss3/accuracy08 = 0.944
I0607 02:09:27.145393 32403 solver.cpp:406] Test net output #107: loss3/accuracy09 = 0.97
I0607 02:09:27.145412 32403 solver.cpp:406] Test net output #108: loss3/accuracy10 = 0.984
I0607 02:09:27.145428 32403 solver.cpp:406] Test net output #109: loss3/accuracy11 = 0.995
I0607 02:09:27.145447 32403 solver.cpp:406] Test net output #110: loss3/accuracy12 = 0.998
I0607 02:09:27.145465 32403 solver.cpp:406] Test net output #111: loss3/accuracy13 = 0.999
I0607 02:09:27.145486 32403 solver.cpp:406] Test net output #112: loss3/accuracy14 = 1
I0607 02:09:27.145505 32403 solver.cpp:406] Test net output #113: loss3/accuracy15 = 1
I0607 02:09:27.145524 32403 solver.cpp:406] Test net output #114: loss3/accuracy16 = 1
I0607 02:09:27.145560 32403 solver.cpp:406] Test net output #115: loss3/accuracy17 = 1
I0607 02:09:27.145582 32403 solver.cpp:406] Test net output #116: loss3/accuracy18 = 1
I0607 02:09:27.145601 32403 solver.cpp:406] Test net output #117: loss3/accuracy19 = 1
I0607 02:09:27.145620 32403 solver.cpp:406] Test net output #118: loss3/accuracy20 = 1
I0607 02:09:27.145640 32403 solver.cpp:406] Test net output #119: loss3/accuracy21 = 1
I0607 02:09:27.145659 32403 solver.cpp:406] Test net output #120: loss3/accuracy22 = 1
I0607 02:09:27.145678 32403 solver.cpp:406] Test net output #121: loss3/accuracy_incl_empty = 0.953636
I0607 02:09:27.145699 32403 solver.cpp:406] Test net output #122: loss3/accuracy_top3 = 0.931265
I0607 02:09:27.145721 32403 solver.cpp:406] Test net output #123: loss3/cross_entropy_loss = 0.669971 (* 1 = 0.669971 loss)
I0607 02:09:27.145745 32403 solver.cpp:406] Test net output #124: loss3/cross_entropy_loss_incl_empty = 0.20281 (* 1 = 0.20281 loss)
I0607 02:09:27.145768 32403 solver.cpp:406] Test net output #125: loss3/loss01 = 0.500083 (* 0.0909091 = 0.0454621 loss)
I0607 02:09:27.145792 32403 solver.cpp:406] Test net output #126: loss3/loss02 = 0.628617 (* 0.0909091 = 0.057147 loss)
I0607 02:09:27.145815 32403 solver.cpp:406] Test net output #127: loss3/loss03 = 0.677759 (* 0.0909091 = 0.0616144 loss)
I0607 02:09:27.145838 32403 solver.cpp:406] Test net output #128: loss3/loss04 = 0.738342 (* 0.0909091 = 0.067122 loss)
I0607 02:09:27.145861 32403 solver.cpp:406] Test net output #129: loss3/loss05 = 0.715656 (* 0.0909091 = 0.0650596 loss)
I0607 02:09:27.145884 32403 solver.cpp:406] Test net output #130: loss3/loss06 = 0.554576 (* 0.0909091 = 0.050416 loss)
I0607 02:09:27.145908 32403 solver.cpp:406] Test net output #131: loss3/loss07 = 0.366442 (* 0.0909091 = 0.0333129 loss)
I0607 02:09:27.145928 32403 solver.cpp:406] Test net output #132: loss3/loss08 = 0.187171 (* 0.0909091 = 0.0170156 loss)
I0607 02:09:27.145954 32403 solver.cpp:406] Test net output #133: loss3/loss09 = 0.115748 (* 0.0909091 = 0.0105225 loss)
I0607 02:09:27.145982 32403 solver.cpp:406] Test net output #134: loss3/loss10 = 0.0701418 (* 0.0909091 = 0.00637653 loss)
I0607 02:09:27.146008 32403 solver.cpp:406] Test net output #135: loss3/loss11 = 0.0187731 (* 0.0909091 = 0.00170665 loss)
I0607 02:09:27.146034 32403 solver.cpp:406] Test net output #136: loss3/loss12 = 0.0077547 (* 0.0909091 = 0.000704973 loss)
I0607 02:09:27.146059 32403 solver.cpp:406] Test net output #137: loss3/loss13 = 0.00343919 (* 0.0909091 = 0.000312653 loss)
I0607 02:09:27.146081 32403 solver.cpp:406] Test net output #138: loss3/loss14 = 0.00130206 (* 0.0909091 = 0.000118369 loss)
I0607 02:09:27.146106 32403 solver.cpp:406] Test net output #139: loss3/loss15 = 0.000497089 (* 0.0909091 = 4.51899e-05 loss)
I0607 02:09:27.146131 32403 solver.cpp:406] Test net output #140: loss3/loss16 = 0.000352281 (* 0.0909091 = 3.20256e-05 loss)
I0607 02:09:27.146155 32403 solver.cpp:406] Test net output #141: loss3/loss17 = 0.000398474 (* 0.0909091 = 3.62249e-05 loss)
I0607 02:09:27.146179 32403 solver.cpp:406] Test net output #142: loss3/loss18 = 0.00045049 (* 0.0909091 = 4.09536e-05 loss)
I0607 02:09:27.146203 32403 solver.cpp:406] Test net output #143: loss3/loss19 = 0.000398042 (* 0.0909091 = 3.61857e-05 loss)
I0607 02:09:27.146227 32403 solver.cpp:406] Test net output #144: loss3/loss20 = 0.000446401 (* 0.0909091 = 4.05819e-05 loss)
I0607 02:09:27.146252 32403 solver.cpp:406] Test net output #145: loss3/loss21 = 0.000470048 (* 0.0909091 = 4.27316e-05 loss)
I0607 02:09:27.146276 32403 solver.cpp:406] Test net output #146: loss3/loss22 = 0.000377899 (* 0.0909091 = 3.43544e-05 loss)
I0607 02:09:27.146299 32403 solver.cpp:406] Test net output #147: total_accuracy = 0.598
I0607 02:09:27.146322 32403 solver.cpp:406] Test net output #148: total_accuracy_not_rec = 0.54
I0607 02:09:27.146340 32403 solver.cpp:406] Test net output #149: total_confidence = 0.474373
I0607 02:09:27.146375 32403 solver.cpp:406] Test net output #150: total_confidence_not_rec = 0.424054
I0607 02:09:27.146397 32403 solver.cpp:338] Iteration 5000, Testing net (#1)
I0607 02:10:25.616394 32403 solver.cpp:393] Test loss: 3.72651
I0607 02:10:25.616541 32403 solver.cpp:406] Test net output #0: loss1/accuracy = 0.561259
I0607 02:10:25.616562 32403 solver.cpp:406] Test net output #1: loss1/accuracy01 = 0.754
I0607 02:10:25.616575 32403 solver.cpp:406] Test net output #2: loss1/accuracy02 = 0.651
I0607 02:10:25.616587 32403 solver.cpp:406] Test net output #3: loss1/accuracy03 = 0.537
I0607 02:10:25.616598 32403 solver.cpp:406] Test net output #4: loss1/accuracy04 = 0.42
I0607 02:10:25.616611 32403 solver.cpp:406] Test net output #5: loss1/accuracy05 = 0.478
I0607 02:10:25.616622 32403 solver.cpp:406] Test net output #6: loss1/accuracy06 = 0.64
I0607 02:10:25.616634 32403 solver.cpp:406] Test net output #7: loss1/accuracy07 = 0.757
I0607 02:10:25.616646 32403 solver.cpp:406] Test net output #8: loss1/accuracy08 = 0.805
I0607 02:10:25.616657 32403 solver.cpp:406] Test net output #9: loss1/accuracy09 = 0.836
I0607 02:10:25.616668 32403 solver.cpp:406] Test net output #10: loss1/accuracy10 = 0.853
I0607 02:10:25.616680 32403 solver.cpp:406] Test net output #11: loss1/accuracy11 = 0.894
I0607 02:10:25.616693 32403 solver.cpp:406] Test net output #12: loss1/accuracy12 = 0.896
I0607 02:10:25.616704 32403 solver.cpp:406] Test net output #13: loss1/accuracy13 = 0.92
I0607 02:10:25.616715 32403 solver.cpp:406] Test net output #14: loss1/accuracy14 = 0.931
I0607 02:10:25.616727 32403 solver.cpp:406] Test net output #15: loss1/accuracy15 = 0.951
I0607 02:10:25.616739 32403 solver.cpp:406] Test net output #16: loss1/accuracy16 = 0.965
I0607 02:10:25.616751 32403 solver.cpp:406] Test net output #17: loss1/accuracy17 = 0.983
I0607 02:10:25.616762 32403 solver.cpp:406] Test net output #18: loss1/accuracy18 = 0.988
I0607 02:10:25.616775 32403 solver.cpp:406] Test net output #19: loss1/accuracy19 = 0.99
I0607 02:10:25.616786 32403 solver.cpp:406] Test net output #20: loss1/accuracy20 = 0.996
I0607 02:10:25.616797 32403 solver.cpp:406] Test net output #21: loss1/accuracy21 = 0.999
I0607 02:10:25.616809 32403 solver.cpp:406] Test net output #22: loss1/accuracy22 = 1
I0607 02:10:25.616821 32403 solver.cpp:406] Test net output #23: loss1/accuracy_incl_empty = 0.836866
I0607 02:10:25.616832 32403 solver.cpp:406] Test net output #24: loss1/accuracy_top3 = 0.789678
I0607 02:10:25.616848 32403 solver.cpp:406] Test net output #25: loss1/cross_entropy_loss = 1.62258 (* 0.3 = 0.486773 loss)
I0607 02:10:25.616863 32403 solver.cpp:406] Test net output #26: loss1/cross_entropy_loss_incl_empty = 0.598509 (* 0.3 = 0.179553 loss)
I0607 02:10:25.616880 32403 solver.cpp:406] Test net output #27: loss1/loss01 = 1.02137 (* 0.0272727 = 0.0278556 loss)
I0607 02:10:25.616894 32403 solver.cpp:406] Test net output #28: loss1/loss02 = 1.42544 (* 0.0272727 = 0.0388757 loss)
I0607 02:10:25.616907 32403 solver.cpp:406] Test net output #29: loss1/loss03 = 1.7831 (* 0.0272727 = 0.0486299 loss)
I0607 02:10:25.616921 32403 solver.cpp:406] Test net output #30: loss1/loss04 = 1.98443 (* 0.0272727 = 0.0541209 loss)
I0607 02:10:25.616935 32403 solver.cpp:406] Test net output #31: loss1/loss05 = 1.75957 (* 0.0272727 = 0.0479882 loss)
I0607 02:10:25.616948 32403 solver.cpp:406] Test net output #32: loss1/loss06 = 1.27706 (* 0.0272727 = 0.0348288 loss)
I0607 02:10:25.616962 32403 solver.cpp:406] Test net output #33: loss1/loss07 = 0.872096 (* 0.0272727 = 0.0237844 loss)
I0607 02:10:25.616976 32403 solver.cpp:406] Test net output #34: loss1/loss08 = 0.708471 (* 0.0272727 = 0.0193219 loss)
I0607 02:10:25.616989 32403 solver.cpp:406] Test net output #35: loss1/loss09 = 0.604061 (* 0.0272727 = 0.0164744 loss)
I0607 02:10:25.617002 32403 solver.cpp:406] Test net output #36: loss1/loss10 = 0.529352 (* 0.0272727 = 0.0144369 loss)
I0607 02:10:25.617017 32403 solver.cpp:406] Test net output #37: loss1/loss11 = 0.463325 (* 0.0272727 = 0.0126361 loss)
I0607 02:10:25.617029 32403 solver.cpp:406] Test net output #38: loss1/loss12 = 0.378337 (* 0.0272727 = 0.0103183 loss)
I0607 02:10:25.617063 32403 solver.cpp:406] Test net output #39: loss1/loss13 = 0.290742 (* 0.0272727 = 0.00792933 loss)
I0607 02:10:25.617079 32403 solver.cpp:406] Test net output #40: loss1/loss14 = 0.251842 (* 0.0272727 = 0.00686843 loss)
I0607 02:10:25.617091 32403 solver.cpp:406] Test net output #41: loss1/loss15 = 0.196191 (* 0.0272727 = 0.00535065 loss)
I0607 02:10:25.617105 32403 solver.cpp:406] Test net output #42: loss1/loss16 = 0.16234 (* 0.0272727 = 0.00442744 loss)
I0607 02:10:25.617131 32403 solver.cpp:406] Test net output #43: loss1/loss17 = 0.0945074 (* 0.0272727 = 0.00257747 loss)
I0607 02:10:25.617148 32403 solver.cpp:406] Test net output #44: loss1/loss18 = 0.0724002 (* 0.0272727 = 0.00197455 loss)
I0607 02:10:25.617162 32403 solver.cpp:406] Test net output #45: loss1/loss19 = 0.062046 (* 0.0272727 = 0.00169216 loss)
I0607 02:10:25.617175 32403 solver.cpp:406] Test net output #46: loss1/loss20 = 0.0290862 (* 0.0272727 = 0.00079326 loss)
I0607 02:10:25.617189 32403 solver.cpp:406] Test net output #47: loss1/loss21 = 0.0103788 (* 0.0272727 = 0.000283058 loss)
I0607 02:10:25.617203 32403 solver.cpp:406] Test net output #48: loss1/loss22 = 0.00360998 (* 0.0272727 = 9.8454e-05 loss)
I0607 02:10:25.617215 32403 solver.cpp:406] Test net output #49: loss2/accuracy = 0.680724
I0607 02:10:25.617228 32403 solver.cpp:406] Test net output #50: loss2/accuracy01 = 0.852
I0607 02:10:25.617239 32403 solver.cpp:406] Test net output #51: loss2/accuracy02 = 0.796
I0607 02:10:25.617250 32403 solver.cpp:406] Test net output #52: loss2/accuracy03 = 0.722
I0607 02:10:25.617262 32403 solver.cpp:406] Test net output #53: loss2/accuracy04 = 0.621
I0607 02:10:25.617274 32403 solver.cpp:406] Test net output #54: loss2/accuracy05 = 0.6
I0607 02:10:25.617285 32403 solver.cpp:406] Test net output #55: loss2/accuracy06 = 0.715
I0607 02:10:25.617296 32403 solver.cpp:406] Test net output #56: loss2/accuracy07 = 0.766
I0607 02:10:25.617307 32403 solver.cpp:406] Test net output #57: loss2/accuracy08 = 0.823
I0607 02:10:25.617319 32403 solver.cpp:406] Test net output #58: loss2/accuracy09 = 0.851
I0607 02:10:25.617331 32403 solver.cpp:406] Test net output #59: loss2/accuracy10 = 0.864
I0607 02:10:25.617342 32403 solver.cpp:406] Test net output #60: loss2/accuracy11 = 0.89
I0607 02:10:25.617353 32403 solver.cpp:406] Test net output #61: loss2/accuracy12 = 0.909
I0607 02:10:25.617364 32403 solver.cpp:406] Test net output #62: loss2/accuracy13 = 0.93
I0607 02:10:25.617377 32403 solver.cpp:406] Test net output #63: loss2/accuracy14 = 0.94
I0607 02:10:25.617388 32403 solver.cpp:406] Test net output #64: loss2/accuracy15 = 0.95
I0607 02:10:25.617399 32403 solver.cpp:406] Test net output #65: loss2/accuracy16 = 0.965
I0607 02:10:25.617410 32403 solver.cpp:406] Test net output #66: loss2/accuracy17 = 0.983
I0607 02:10:25.617422 32403 solver.cpp:406] Test net output #67: loss2/accuracy18 = 0.988
I0607 02:10:25.617434 32403 solver.cpp:406] Test net output #68: loss2/accuracy19 = 0.99
I0607 02:10:25.617444 32403 solver.cpp:406] Test net output #69: loss2/accuracy20 = 0.996
I0607 02:10:25.617456 32403 solver.cpp:406] Test net output #70: loss2/accuracy21 = 0.999
I0607 02:10:25.617467 32403 solver.cpp:406] Test net output #71: loss2/accuracy22 = 1
I0607 02:10:25.617478 32403 solver.cpp:406] Test net output #72: loss2/accuracy_incl_empty = 0.87541
I0607 02:10:25.617489 32403 solver.cpp:406] Test net output #73: loss2/accuracy_top3 = 0.858342
I0607 02:10:25.617504 32403 solver.cpp:406] Test net output #74: loss2/cross_entropy_loss = 1.24108 (* 0.3 = 0.372323 loss)
I0607 02:10:25.617514 32403 solver.cpp:406] Test net output #75: loss2/cross_entropy_loss_incl_empty = 0.466965 (* 0.3 = 0.140089 loss)
I0607 02:10:25.617528 32403 solver.cpp:406] Test net output #76: loss2/loss01 = 0.755045 (* 0.0272727 = 0.0205921 loss)
I0607 02:10:25.617542 32403 solver.cpp:406] Test net output #77: loss2/loss02 = 0.958031 (* 0.0272727 = 0.0261281 loss)
I0607 02:10:25.617570 32403 solver.cpp:406] Test net output #78: loss2/loss03 = 1.14995 (* 0.0272727 = 0.0313621 loss)
I0607 02:10:25.617585 32403 solver.cpp:406] Test net output #79: loss2/loss04 = 1.35498 (* 0.0272727 = 0.0369541 loss)
I0607 02:10:25.617599 32403 solver.cpp:406] Test net output #80: loss2/loss05 = 1.33685 (* 0.0272727 = 0.0364595 loss)
I0607 02:10:25.617616 32403 solver.cpp:406] Test net output #81: loss2/loss06 = 0.988659 (* 0.0272727 = 0.0269634 loss)
I0607 02:10:25.617642 32403 solver.cpp:406] Test net output #82: loss2/loss07 = 0.793691 (* 0.0272727 = 0.0216461 loss)
I0607 02:10:25.617666 32403 solver.cpp:406] Test net output #83: loss2/loss08 = 0.630246 (* 0.0272727 = 0.0171885 loss)
I0607 02:10:25.617679 32403 solver.cpp:406] Test net output #84: loss2/loss09 = 0.553406 (* 0.0272727 = 0.0150929 loss)
I0607 02:10:25.617693 32403 solver.cpp:406] Test net output #85: loss2/loss10 = 0.476464 (* 0.0272727 = 0.0129945 loss)
I0607 02:10:25.617707 32403 solver.cpp:406] Test net output #86: loss2/loss11 = 0.429001 (* 0.0272727 = 0.0117 loss)
I0607 02:10:25.617720 32403 solver.cpp:406] Test net output #87: loss2/loss12 = 0.329958 (* 0.0272727 = 0.00899887 loss)
I0607 02:10:25.617734 32403 solver.cpp:406] Test net output #88: loss2/loss13 = 0.251042 (* 0.0272727 = 0.0068466 loss)
I0607 02:10:25.617748 32403 solver.cpp:406] Test net output #89: loss2/loss14 = 0.216012 (* 0.0272727 = 0.00589123 loss)
I0607 02:10:25.617761 32403 solver.cpp:406] Test net output #90: loss2/loss15 = 0.176095 (* 0.0272727 = 0.0048026 loss)
I0607 02:10:25.617774 32403 solver.cpp:406] Test net output #91: loss2/loss16 = 0.146254 (* 0.0272727 = 0.00398876 loss)
I0607 02:10:25.617789 32403 solver.cpp:406] Test net output #92: loss2/loss17 = 0.0826619 (* 0.0272727 = 0.00225442 loss)
I0607 02:10:25.617802 32403 solver.cpp:406] Test net output #93: loss2/loss18 = 0.0637381 (* 0.0272727 = 0.00173831 loss)
I0607 02:10:25.617816 32403 solver.cpp:406] Test net output #94: loss2/loss19 = 0.0614346 (* 0.0272727 = 0.00167549 loss)
I0607 02:10:25.617830 32403 solver.cpp:406] Test net output #95: loss2/loss20 = 0.0269692 (* 0.0272727 = 0.000735524 loss)
I0607 02:10:25.617843 32403 solver.cpp:406] Test net output #96: loss2/loss21 = 0.00822977 (* 0.0272727 = 0.000224448 loss)
I0607 02:10:25.617857 32403 solver.cpp:406] Test net output #97: loss2/loss22 = 0.00136776 (* 0.0272727 = 3.73025e-05 loss)
I0607 02:10:25.617868 32403 solver.cpp:406] Test net output #98: loss3/accuracy = 0.800907
I0607 02:10:25.617880 32403 solver.cpp:406] Test net output #99: loss3/accuracy01 = 0.876
I0607 02:10:25.617892 32403 solver.cpp:406] Test net output #100: loss3/accuracy02 = 0.845
I0607 02:10:25.617903 32403 solver.cpp:406] Test net output #101: loss3/accuracy03 = 0.825
I0607 02:10:25.617914 32403 solver.cpp:406] Test net output #102: loss3/accuracy04 = 0.829
I0607 02:10:25.617928 32403 solver.cpp:406] Test net output #103: loss3/accuracy05 = 0.824
I0607 02:10:25.617940 32403 solver.cpp:406] Test net output #104: loss3/accuracy06 = 0.849
I0607 02:10:25.617951 32403 solver.cpp:406] Test net output #105: loss3/accuracy07 = 0.854
I0607 02:10:25.617962 32403 solver.cpp:406] Test net output #106: loss3/accuracy08 = 0.874
I0607 02:10:25.617974 32403 solver.cpp:406] Test net output #107: loss3/accuracy09 = 0.889
I0607 02:10:25.617985 32403 solver.cpp:406] Test net output #108: loss3/accuracy10 = 0.901
I0607 02:10:25.617996 32403 solver.cpp:406] Test net output #109: loss3/accuracy11 = 0.923
I0607 02:10:25.618007 32403 solver.cpp:406] Test net output #110: loss3/accuracy12 = 0.923
I0607 02:10:25.618018 32403 solver.cpp:406] Test net output #111: loss3/accuracy13 = 0.938
I0607 02:10:25.618031 32403 solver.cpp:406] Test net output #112: loss3/accuracy14 = 0.955
I0607 02:10:25.618041 32403 solver.cpp:406] Test net output #113: loss3/accuracy15 = 0.964
I0607 02:10:25.618052 32403 solver.cpp:406] Test net output #114: loss3/accuracy16 = 0.97
I0607 02:10:25.618075 32403 solver.cpp:406] Test net output #115: loss3/accuracy17 = 0.985
I0607 02:10:25.618088 32403 solver.cpp:406] Test net output #116: loss3/accuracy18 = 0.989
I0607 02:10:25.618099 32403 solver.cpp:406] Test net output #117: loss3/accuracy19 = 0.99
I0607 02:10:25.618110 32403 solver.cpp:406] Test net output #118: loss3/accuracy20 = 0.996
I0607 02:10:25.618121 32403 solver.cpp:406] Test net output #119: loss3/accuracy21 = 0.999
I0607 02:10:25.618132 32403 solver.cpp:406] Test net output #120: loss3/accuracy22 = 1
I0607 02:10:25.618144 32403 solver.cpp:406] Test net output #121: loss3/accuracy_incl_empty = 0.920137
I0607 02:10:25.618155 32403 solver.cpp:406] Test net output #122: loss3/accuracy_top3 = 0.905087
I0607 02:10:25.618168 32403 solver.cpp:406] Test net output #123: loss3/cross_entropy_loss = 0.866201 (* 1 = 0.866201 loss)
I0607 02:10:25.618181 32403 solver.cpp:406] Test net output #124: loss3/cross_entropy_loss_incl_empty = 0.325578 (* 1 = 0.325578 loss)
I0607 02:10:25.618196 32403 solver.cpp:406] Test net output #125: loss3/loss01 = 0.618755 (* 0.0909091 = 0.0562505 loss)
I0607 02:10:25.618208 32403 solver.cpp:406] Test net output #126: loss3/loss02 = 0.772403 (* 0.0909091 = 0.0702185 loss)
I0607 02:10:25.618221 32403 solver.cpp:406] Test net output #127: loss3/loss03 = 0.829488 (* 0.0909091 = 0.075408 loss)
I0607 02:10:25.618235 32403 solver.cpp:406] Test net output #128: loss3/loss04 = 0.779997 (* 0.0909091 = 0.0709088 loss)
I0607 02:10:25.618245 32403 solver.cpp:406] Test net output #129: loss3/loss05 = 0.79317 (* 0.0909091 = 0.0721064 loss)
I0607 02:10:25.618260 32403 solver.cpp:406] Test net output #130: loss3/loss06 = 0.643741 (* 0.0909091 = 0.0585219 loss)
I0607 02:10:25.618274 32403 solver.cpp:406] Test net output #131: loss3/loss07 = 0.523633 (* 0.0909091 = 0.047603 loss)
I0607 02:10:25.618288 32403 solver.cpp:406] Test net output #132: loss3/loss08 = 0.472454 (* 0.0909091 = 0.0429504 loss)
I0607 02:10:25.618301 32403 solver.cpp:406] Test net output #133: loss3/loss09 = 0.387691 (* 0.0909091 = 0.0352446 loss)
I0607 02:10:25.618314 32403 solver.cpp:406] Test net output #134: loss3/loss10 = 0.355836 (* 0.0909091 = 0.0323487 loss)
I0607 02:10:25.618327 32403 solver.cpp:406] Test net output #135: loss3/loss11 = 0.3013 (* 0.0909091 = 0.0273909 loss)
I0607 02:10:25.618341 32403 solver.cpp:406] Test net output #136: loss3/loss12 = 0.257372 (* 0.0909091 = 0.0233975 loss)
I0607 02:10:25.618355 32403 solver.cpp:406] Test net output #137: loss3/loss13 = 0.197598 (* 0.0909091 = 0.0179635 loss)
I0607 02:10:25.618367 32403 solver.cpp:406] Test net output #138: loss3/loss14 = 0.151075 (* 0.0909091 = 0.0137341 loss)
I0607 02:10:25.618381 32403 solver.cpp:406] Test net output #139: loss3/loss15 = 0.125959 (* 0.0909091 = 0.0114508 loss)
I0607 02:10:25.618394 32403 solver.cpp:406] Test net output #140: loss3/loss16 = 0.103122 (* 0.0909091 = 0.00937475 loss)
I0607 02:10:25.618408 32403 solver.cpp:406] Test net output #141: loss3/loss17 = 0.0605599 (* 0.0909091 = 0.00550545 loss)
I0607 02:10:25.618422 32403 solver.cpp:406] Test net output #142: loss3/loss18 = 0.0435105 (* 0.0909091 = 0.0039555 loss)
I0607 02:10:25.618434 32403 solver.cpp:406] Test net output #143: loss3/loss19 = 0.0426461 (* 0.0909091 = 0.00387691 loss)
I0607 02:10:25.618448 32403 solver.cpp:406] Test net output #144: loss3/loss20 = 0.0176975 (* 0.0909091 = 0.00160886 loss)
I0607 02:10:25.618461 32403 solver.cpp:406] Test net output #145: loss3/loss21 = 0.00593004 (* 0.0909091 = 0.000539095 loss)
I0607 02:10:25.618474 32403 solver.cpp:406] Test net output #146: loss3/loss22 = 0.00101892 (* 0.0909091 = 9.2629e-05 loss)
I0607 02:10:25.618486 32403 solver.cpp:406] Test net output #147: total_accuracy = 0.483
I0607 02:10:25.618497 32403 solver.cpp:406] Test net output #148: total_accuracy_not_rec = 0.449
I0607 02:10:25.618508 32403 solver.cpp:406] Test net output #149: total_confidence = 0.372669
I0607 02:10:25.618528 32403 solver.cpp:406] Test net output #150: total_confidence_not_rec = 0.330825
I0607 02:10:25.976158 32403 solver.cpp:229] Iteration 5000, loss = 4.11076
I0607 02:10:25.976215 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.5
I0607 02:10:25.976233 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.875
I0607 02:10:25.976246 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.625
I0607 02:10:25.976258 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.25
I0607 02:10:25.976271 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.5
I0607 02:10:25.976284 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.375
I0607 02:10:25.976295 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.5
I0607 02:10:25.976307 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.75
I0607 02:10:25.976320 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 1
I0607 02:10:25.976331 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.875
I0607 02:10:25.976343 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 1
I0607 02:10:25.976356 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.875
I0607 02:10:25.976367 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 02:10:25.976379 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.875
I0607 02:10:25.976392 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.875
I0607 02:10:25.976403 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 0.875
I0607 02:10:25.976415 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 02:10:25.976428 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 02:10:25.976439 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 02:10:25.976451 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 02:10:25.976464 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 02:10:25.976475 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 02:10:25.976487 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 02:10:25.976498 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.840909
I0607 02:10:25.976511 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.740741
I0607 02:10:25.976527 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.60472 (* 0.3 = 0.481415 loss)
I0607 02:10:25.976541 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.524763 (* 0.3 = 0.157429 loss)
I0607 02:10:25.976557 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.436755 (* 0.0272727 = 0.0119115 loss)
I0607 02:10:25.976570 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.49836 (* 0.0272727 = 0.0408645 loss)
I0607 02:10:25.976584 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.72562 (* 0.0272727 = 0.0470623 loss)
I0607 02:10:25.976598 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.67843 (* 0.0272727 = 0.0457753 loss)
I0607 02:10:25.976611 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.60085 (* 0.0272727 = 0.0436594 loss)
I0607 02:10:25.976625 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.8907 (* 0.0272727 = 0.0515645 loss)
I0607 02:10:25.976639 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 0.758076 (* 0.0272727 = 0.0206748 loss)
I0607 02:10:25.976654 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.24864 (* 0.0272727 = 0.0067811 loss)
I0607 02:10:25.976666 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.382349 (* 0.0272727 = 0.0104277 loss)
I0607 02:10:25.976683 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.265304 (* 0.0272727 = 0.00723556 loss)
I0607 02:10:25.976706 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.411561 (* 0.0272727 = 0.0112244 loss)
I0607 02:10:25.976764 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.408272 (* 0.0272727 = 0.0111347 loss)
I0607 02:10:25.976781 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.449716 (* 0.0272727 = 0.012265 loss)
I0607 02:10:25.976795 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.414575 (* 0.0272727 = 0.0113066 loss)
I0607 02:10:25.976810 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.38952 (* 0.0272727 = 0.0106233 loss)
I0607 02:10:25.976824 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.121868 (* 0.0272727 = 0.00332367 loss)
I0607 02:10:25.976838 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.0582917 (* 0.0272727 = 0.00158977 loss)
I0607 02:10:25.976852 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.0416189 (* 0.0272727 = 0.00113506 loss)
I0607 02:10:25.976866 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.0248354 (* 0.0272727 = 0.000677329 loss)
I0607 02:10:25.976881 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.0332815 (* 0.0272727 = 0.000907676 loss)
I0607 02:10:25.976894 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.0183761 (* 0.0272727 = 0.000501166 loss)
I0607 02:10:25.976909 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.0096825 (* 0.0272727 = 0.000264068 loss)
I0607 02:10:25.976922 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.666667
I0607 02:10:25.976933 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 1
I0607 02:10:25.976945 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 1
I0607 02:10:25.976956 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.75
I0607 02:10:25.976969 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.375
I0607 02:10:25.976980 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.625
I0607 02:10:25.976992 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.625
I0607 02:10:25.977007 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.75
I0607 02:10:25.977020 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.875
I0607 02:10:25.977031 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.875
I0607 02:10:25.977043 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 1
I0607 02:10:25.977056 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.875
I0607 02:10:25.977066 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.875
I0607 02:10:25.977078 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.875
I0607 02:10:25.977090 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.875
I0607 02:10:25.977102 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 0.875
I0607 02:10:25.977114 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 02:10:25.977143 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 02:10:25.977155 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 02:10:25.977166 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 02:10:25.977179 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 02:10:25.977190 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 02:10:25.977201 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 02:10:25.977212 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.892045
I0607 02:10:25.977224 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.796296
I0607 02:10:25.977238 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.03271 (* 0.3 = 0.309812 loss)
I0607 02:10:25.977252 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.340674 (* 0.3 = 0.102202 loss)
I0607 02:10:25.977279 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.140528 (* 0.0272727 = 0.00383257 loss)
I0607 02:10:25.977294 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.265936 (* 0.0272727 = 0.00725281 loss)
I0607 02:10:25.977308 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.672438 (* 0.0272727 = 0.0183392 loss)
I0607 02:10:25.977322 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.48195 (* 0.0272727 = 0.0404167 loss)
I0607 02:10:25.977336 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.22268 (* 0.0272727 = 0.0333458 loss)
I0607 02:10:25.977349 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.52627 (* 0.0272727 = 0.0416255 loss)
I0607 02:10:25.977363 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.657367 (* 0.0272727 = 0.0179282 loss)
I0607 02:10:25.977377 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.291782 (* 0.0272727 = 0.00795768 loss)
I0607 02:10:25.977391 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.25125 (* 0.0272727 = 0.00685228 loss)
I0607 02:10:25.977404 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.192214 (* 0.0272727 = 0.00524221 loss)
I0607 02:10:25.977418 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.407388 (* 0.0272727 = 0.0111106 loss)
I0607 02:10:25.977432 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.4921 (* 0.0272727 = 0.0134209 loss)
I0607 02:10:25.977447 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.549603 (* 0.0272727 = 0.0149892 loss)
I0607 02:10:25.977460 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.516216 (* 0.0272727 = 0.0140786 loss)
I0607 02:10:25.977474 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.364185 (* 0.0272727 = 0.00993233 loss)
I0607 02:10:25.977488 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.0373006 (* 0.0272727 = 0.00101729 loss)
I0607 02:10:25.977502 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.0102873 (* 0.0272727 = 0.000280563 loss)
I0607 02:10:25.977516 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.00450612 (* 0.0272727 = 0.000122894 loss)
I0607 02:10:25.977531 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.00323545 (* 0.0272727 = 8.82396e-05 loss)
I0607 02:10:25.977545 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.000904863 (* 0.0272727 = 2.46781e-05 loss)
I0607 02:10:25.977558 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.0014362 (* 0.0272727 = 3.9169e-05 loss)
I0607 02:10:25.977573 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 0.000988623 (* 0.0272727 = 2.69625e-05 loss)
I0607 02:10:25.977586 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.777778
I0607 02:10:25.977604 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 02:10:25.977624 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 1
I0607 02:10:25.977638 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 1
I0607 02:10:25.977649 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.75
I0607 02:10:25.977660 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.75
I0607 02:10:25.977672 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.75
I0607 02:10:25.977680 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.875
I0607 02:10:25.977689 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 1
I0607 02:10:25.977700 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.875
I0607 02:10:25.977711 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.875
I0607 02:10:25.977723 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.875
I0607 02:10:25.977737 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.875
I0607 02:10:25.977749 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 0.875
I0607 02:10:25.977771 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 0.875
I0607 02:10:25.977784 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 0.875
I0607 02:10:25.977797 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 02:10:25.977808 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 02:10:25.977819 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 02:10:25.977831 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 02:10:25.977843 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 02:10:25.977854 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 02:10:25.977865 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 02:10:25.977876 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.931818
I0607 02:10:25.977888 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.907407
I0607 02:10:25.977902 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.661651 (* 1 = 0.661651 loss)
I0607 02:10:25.977916 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.211358 (* 1 = 0.211358 loss)
I0607 02:10:25.977931 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.079143 (* 0.0909091 = 0.00719482 loss)
I0607 02:10:25.977946 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.111446 (* 0.0909091 = 0.0101314 loss)
I0607 02:10:25.977959 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.235656 (* 0.0909091 = 0.0214233 loss)
I0607 02:10:25.977973 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.461335 (* 0.0909091 = 0.0419395 loss)
I0607 02:10:25.977987 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.491505 (* 0.0909091 = 0.0446823 loss)
I0607 02:10:25.978000 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.654604 (* 0.0909091 = 0.0595094 loss)
I0607 02:10:25.978014 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.39611 (* 0.0909091 = 0.03601 loss)
I0607 02:10:25.978029 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.0944048 (* 0.0909091 = 0.00858226 loss)
I0607 02:10:25.978042 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.357713 (* 0.0909091 = 0.0325194 loss)
I0607 02:10:25.978060 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.264051 (* 0.0909091 = 0.0240047 loss)
I0607 02:10:25.978075 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.402949 (* 0.0909091 = 0.0366318 loss)
I0607 02:10:25.978090 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.342766 (* 0.0909091 = 0.0311605 loss)
I0607 02:10:25.978102 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.427919 (* 0.0909091 = 0.0389017 loss)
I0607 02:10:25.978116 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.354319 (* 0.0909091 = 0.0322108 loss)
I0607 02:10:25.978129 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.289561 (* 0.0909091 = 0.0263237 loss)
I0607 02:10:25.978143 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.0136072 (* 0.0909091 = 0.00123702 loss)
I0607 02:10:25.978157 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.00351688 (* 0.0909091 = 0.000319717 loss)
I0607 02:10:25.978171 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.00145465 (* 0.0909091 = 0.000132241 loss)
I0607 02:10:25.978185 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.000472674 (* 0.0909091 = 4.29704e-05 loss)
I0607 02:10:25.978199 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.000508548 (* 0.0909091 = 4.62317e-05 loss)
I0607 02:10:25.978212 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.000598008 (* 0.0909091 = 5.43644e-05 loss)
I0607 02:10:25.978226 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 0.000181065 (* 0.0909091 = 1.64604e-05 loss)
I0607 02:10:25.978248 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.625
I0607 02:10:25.978261 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.625
I0607 02:10:25.978273 32403 solver.cpp:245] Train net output #149: total_confidence = 0.529999
I0607 02:10:25.978286 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.498174
I0607 02:10:25.978298 32403 sgd_solver.cpp:106] Iteration 5000, lr = 0.001
I0607 02:14:49.723337 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.0256 > 30) by scale factor 0.881688
I0607 02:16:53.380003 32403 solver.cpp:229] Iteration 5500, loss = 4.09199
I0607 02:16:53.380151 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.483871
I0607 02:16:53.380172 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.75
I0607 02:16:53.380187 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.625
I0607 02:16:53.380199 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.625
I0607 02:16:53.380213 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.625
I0607 02:16:53.380224 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.5
I0607 02:16:53.380236 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.5
I0607 02:16:53.380249 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.75
I0607 02:16:53.380261 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.75
I0607 02:16:53.380273 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.625
I0607 02:16:53.380286 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.75
I0607 02:16:53.380298 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.75
I0607 02:16:53.380311 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 02:16:53.380322 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 1
I0607 02:16:53.380334 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.875
I0607 02:16:53.380347 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 0.875
I0607 02:16:53.380358 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 02:16:53.380370 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 02:16:53.380383 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 02:16:53.380394 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 02:16:53.380405 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 02:16:53.380419 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 02:16:53.380429 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 02:16:53.380441 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.8125
I0607 02:16:53.380455 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.677419
I0607 02:16:53.380471 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.74143 (* 0.3 = 0.522429 loss)
I0607 02:16:53.380486 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.665689 (* 0.3 = 0.199707 loss)
I0607 02:16:53.380501 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.752636 (* 0.0272727 = 0.0205264 loss)
I0607 02:16:53.380516 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.20407 (* 0.0272727 = 0.0328382 loss)
I0607 02:16:53.380530 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.30765 (* 0.0272727 = 0.0356631 loss)
I0607 02:16:53.380544 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.0669 (* 0.0272727 = 0.0290974 loss)
I0607 02:16:53.380558 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.89419 (* 0.0272727 = 0.0516598 loss)
I0607 02:16:53.380571 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.42807 (* 0.0272727 = 0.0389473 loss)
I0607 02:16:53.380585 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 1.1074 (* 0.0272727 = 0.0302019 loss)
I0607 02:16:53.380600 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.819045 (* 0.0272727 = 0.0223376 loss)
I0607 02:16:53.380614 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 1.99383 (* 0.0272727 = 0.0543772 loss)
I0607 02:16:53.380628 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.649606 (* 0.0272727 = 0.0177165 loss)
I0607 02:16:53.380642 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.61403 (* 0.0272727 = 0.0167463 loss)
I0607 02:16:53.380657 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.351478 (* 0.0272727 = 0.00958576 loss)
I0607 02:16:53.380692 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.28168 (* 0.0272727 = 0.00768219 loss)
I0607 02:16:53.380708 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.284222 (* 0.0272727 = 0.00775152 loss)
I0607 02:16:53.380723 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.273749 (* 0.0272727 = 0.00746587 loss)
I0607 02:16:53.380738 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.0498511 (* 0.0272727 = 0.00135958 loss)
I0607 02:16:53.380753 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.000618739 (* 0.0272727 = 1.68747e-05 loss)
I0607 02:16:53.380766 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.000690816 (* 0.0272727 = 1.88404e-05 loss)
I0607 02:16:53.380780 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.000274183 (* 0.0272727 = 7.47772e-06 loss)
I0607 02:16:53.380795 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.000375258 (* 0.0272727 = 1.02343e-05 loss)
I0607 02:16:53.380808 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 6.7681e-05 (* 0.0272727 = 1.84585e-06 loss)
I0607 02:16:53.380823 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.000117762 (* 0.0272727 = 3.21169e-06 loss)
I0607 02:16:53.380836 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.548387
I0607 02:16:53.380847 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.875
I0607 02:16:53.380861 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.75
I0607 02:16:53.380872 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.5
I0607 02:16:53.380888 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.75
I0607 02:16:53.380900 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.5
I0607 02:16:53.380913 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.5
I0607 02:16:53.380924 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.75
I0607 02:16:53.380936 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.875
I0607 02:16:53.380949 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.625
I0607 02:16:53.380960 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.75
I0607 02:16:53.380972 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.75
I0607 02:16:53.380985 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.875
I0607 02:16:53.380997 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.875
I0607 02:16:53.381009 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 1
I0607 02:16:53.381021 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 0.875
I0607 02:16:53.381033 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 02:16:53.381045 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 02:16:53.381057 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 02:16:53.381068 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 02:16:53.381079 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 02:16:53.381091 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 02:16:53.381103 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 02:16:53.381114 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.823864
I0607 02:16:53.381141 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.790323
I0607 02:16:53.381157 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.43223 (* 0.3 = 0.42967 loss)
I0607 02:16:53.381171 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.56983 (* 0.3 = 0.170949 loss)
I0607 02:16:53.381186 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.386332 (* 0.0272727 = 0.0105363 loss)
I0607 02:16:53.381199 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 1.36058 (* 0.0272727 = 0.0371068 loss)
I0607 02:16:53.381227 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 1.40315 (* 0.0272727 = 0.0382677 loss)
I0607 02:16:53.381242 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 0.806648 (* 0.0272727 = 0.0219995 loss)
I0607 02:16:53.381255 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.74233 (* 0.0272727 = 0.0475181 loss)
I0607 02:16:53.381269 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.40195 (* 0.0272727 = 0.038235 loss)
I0607 02:16:53.381283 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 1.17646 (* 0.0272727 = 0.0320852 loss)
I0607 02:16:53.381297 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.855853 (* 0.0272727 = 0.0233414 loss)
I0607 02:16:53.381311 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 1.30728 (* 0.0272727 = 0.0356531 loss)
I0607 02:16:53.381325 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.570816 (* 0.0272727 = 0.0155677 loss)
I0607 02:16:53.381340 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.75399 (* 0.0272727 = 0.0205634 loss)
I0607 02:16:53.381353 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.539526 (* 0.0272727 = 0.0147144 loss)
I0607 02:16:53.381368 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.232057 (* 0.0272727 = 0.00632882 loss)
I0607 02:16:53.381382 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.164717 (* 0.0272727 = 0.00449228 loss)
I0607 02:16:53.381397 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.257523 (* 0.0272727 = 0.00702335 loss)
I0607 02:16:53.381412 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.0490344 (* 0.0272727 = 0.0013373 loss)
I0607 02:16:53.381425 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.00166767 (* 0.0272727 = 4.5482e-05 loss)
I0607 02:16:53.381439 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.000503666 (* 0.0272727 = 1.37363e-05 loss)
I0607 02:16:53.381454 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.000445311 (* 0.0272727 = 1.21449e-05 loss)
I0607 02:16:53.381469 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.000218962 (* 0.0272727 = 5.9717e-06 loss)
I0607 02:16:53.381482 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.000299114 (* 0.0272727 = 8.15765e-06 loss)
I0607 02:16:53.381496 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 9.07277e-05 (* 0.0272727 = 2.47439e-06 loss)
I0607 02:16:53.381510 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.822581
I0607 02:16:53.381521 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 02:16:53.381533 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.875
I0607 02:16:53.381544 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.875
I0607 02:16:53.381556 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.875
I0607 02:16:53.381568 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 1
I0607 02:16:53.381580 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.875
I0607 02:16:53.381592 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.875
I0607 02:16:53.381603 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.75
I0607 02:16:53.381615 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.75
I0607 02:16:53.381628 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.875
I0607 02:16:53.381639 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.875
I0607 02:16:53.381650 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.75
I0607 02:16:53.381662 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 0.75
I0607 02:16:53.381674 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 0.875
I0607 02:16:53.381685 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 0.875
I0607 02:16:53.381707 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 02:16:53.381721 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 02:16:53.381729 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 02:16:53.381736 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 02:16:53.381744 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 02:16:53.381757 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 02:16:53.381768 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 02:16:53.381780 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.920455
I0607 02:16:53.381793 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.919355
I0607 02:16:53.381806 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.602986 (* 1 = 0.602986 loss)
I0607 02:16:53.381820 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.266876 (* 1 = 0.266876 loss)
I0607 02:16:53.381834 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.177938 (* 0.0909091 = 0.0161761 loss)
I0607 02:16:53.381849 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.58732 (* 0.0909091 = 0.0533928 loss)
I0607 02:16:53.381862 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.310702 (* 0.0909091 = 0.0282456 loss)
I0607 02:16:53.381876 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.240765 (* 0.0909091 = 0.0218877 loss)
I0607 02:16:53.381891 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.200749 (* 0.0909091 = 0.0182499 loss)
I0607 02:16:53.381904 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.543479 (* 0.0909091 = 0.0494072 loss)
I0607 02:16:53.381918 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.507539 (* 0.0909091 = 0.0461399 loss)
I0607 02:16:53.381935 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.822179 (* 0.0909091 = 0.0747435 loss)
I0607 02:16:53.381950 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.957361 (* 0.0909091 = 0.0870329 loss)
I0607 02:16:53.381964 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.345494 (* 0.0909091 = 0.0314086 loss)
I0607 02:16:53.381978 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.397853 (* 0.0909091 = 0.0361685 loss)
I0607 02:16:53.381991 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.317417 (* 0.0909091 = 0.0288561 loss)
I0607 02:16:53.382005 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.363392 (* 0.0909091 = 0.0330356 loss)
I0607 02:16:53.382019 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.277529 (* 0.0909091 = 0.0252299 loss)
I0607 02:16:53.382033 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.248824 (* 0.0909091 = 0.0226203 loss)
I0607 02:16:53.382047 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.0766387 (* 0.0909091 = 0.00696715 loss)
I0607 02:16:53.382061 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.00154902 (* 0.0909091 = 0.00014082 loss)
I0607 02:16:53.382076 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.000668433 (* 0.0909091 = 6.07667e-05 loss)
I0607 02:16:53.382089 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.000227285 (* 0.0909091 = 2.06623e-05 loss)
I0607 02:16:53.382103 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.000136639 (* 0.0909091 = 1.24217e-05 loss)
I0607 02:16:53.382117 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.000191911 (* 0.0909091 = 1.74465e-05 loss)
I0607 02:16:53.382131 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 0.000105756 (* 0.0909091 = 9.61415e-06 loss)
I0607 02:16:53.382143 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.625
I0607 02:16:53.382155 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.625
I0607 02:16:53.382179 32403 solver.cpp:245] Train net output #149: total_confidence = 0.522011
I0607 02:16:53.382194 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.47312
I0607 02:16:53.382206 32403 sgd_solver.cpp:106] Iteration 5500, lr = 0.001
I0607 02:18:03.478826 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.7248 > 30) by scale factor 0.863937
I0607 02:18:56.908107 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 35.8871 > 30) by scale factor 0.835955
I0607 02:21:23.940065 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 43.4768 > 30) by scale factor 0.690023
I0607 02:22:21.213745 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 36.3388 > 30) by scale factor 0.825564
I0607 02:22:32.818114 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.1251 > 30) by scale factor 0.995846
I0607 02:22:52.934404 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 38.1345 > 30) by scale factor 0.786689
I0607 02:23:20.440461 32403 solver.cpp:229] Iteration 6000, loss = 4.08246
I0607 02:23:20.440536 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.738095
I0607 02:23:20.440556 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.75
I0607 02:23:20.440569 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.75
I0607 02:23:20.440583 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.875
I0607 02:23:20.440595 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.75
I0607 02:23:20.440608 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.5
I0607 02:23:20.440619 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.75
I0607 02:23:20.440631 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.875
I0607 02:23:20.440644 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 1
I0607 02:23:20.440656 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 1
I0607 02:23:20.440668 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 1
I0607 02:23:20.440680 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 1
I0607 02:23:20.440691 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 1
I0607 02:23:20.440703 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 1
I0607 02:23:20.440716 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 1
I0607 02:23:20.440728 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 02:23:20.440740 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 02:23:20.440754 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 02:23:20.440768 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 02:23:20.440779 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 02:23:20.440790 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 02:23:20.440803 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 02:23:20.440814 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 02:23:20.440826 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.931818
I0607 02:23:20.440839 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.904762
I0607 02:23:20.440855 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 0.889205 (* 0.3 = 0.266761 loss)
I0607 02:23:20.440870 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.239976 (* 0.3 = 0.0719929 loss)
I0607 02:23:20.440884 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.412606 (* 0.0272727 = 0.0112529 loss)
I0607 02:23:20.440898 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.10524 (* 0.0272727 = 0.030143 loss)
I0607 02:23:20.440912 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 0.66991 (* 0.0272727 = 0.0182703 loss)
I0607 02:23:20.440927 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 0.942501 (* 0.0272727 = 0.0257046 loss)
I0607 02:23:20.440940 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.21904 (* 0.0272727 = 0.0332466 loss)
I0607 02:23:20.440954 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 0.819711 (* 0.0272727 = 0.0223558 loss)
I0607 02:23:20.440969 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 0.461083 (* 0.0272727 = 0.012575 loss)
I0607 02:23:20.440984 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.0238713 (* 0.0272727 = 0.000651034 loss)
I0607 02:23:20.440999 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.0051927 (* 0.0272727 = 0.000141619 loss)
I0607 02:23:20.441012 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.00224272 (* 0.0272727 = 6.11651e-05 loss)
I0607 02:23:20.441027 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.00144564 (* 0.0272727 = 3.94266e-05 loss)
I0607 02:23:20.441041 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.000184518 (* 0.0272727 = 5.03232e-06 loss)
I0607 02:23:20.441103 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.000229631 (* 0.0272727 = 6.26268e-06 loss)
I0607 02:23:20.441133 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.000114763 (* 0.0272727 = 3.12991e-06 loss)
I0607 02:23:20.441149 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 3.4179e-05 (* 0.0272727 = 9.32155e-07 loss)
I0607 02:23:20.441164 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 2.81201e-05 (* 0.0272727 = 7.66912e-07 loss)
I0607 02:23:20.441179 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 1.30389e-05 (* 0.0272727 = 3.55605e-07 loss)
I0607 02:23:20.441192 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 5.31977e-06 (* 0.0272727 = 1.45085e-07 loss)
I0607 02:23:20.441207 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 4.61943e-06 (* 0.0272727 = 1.25984e-07 loss)
I0607 02:23:20.441221 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 1.91488e-05 (* 0.0272727 = 5.2224e-07 loss)
I0607 02:23:20.441236 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 3.44219e-06 (* 0.0272727 = 9.3878e-08 loss)
I0607 02:23:20.441251 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 1.84775e-06 (* 0.0272727 = 5.03932e-08 loss)
I0607 02:23:20.441262 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.857143
I0607 02:23:20.441275 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 1
I0607 02:23:20.441287 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 1
I0607 02:23:20.441303 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 1
I0607 02:23:20.441314 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.75
I0607 02:23:20.441326 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.5
I0607 02:23:20.441339 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.75
I0607 02:23:20.441351 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.875
I0607 02:23:20.441362 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 1
I0607 02:23:20.441375 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 1
I0607 02:23:20.441386 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 1
I0607 02:23:20.441397 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 1
I0607 02:23:20.441409 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 1
I0607 02:23:20.441421 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 1
I0607 02:23:20.441432 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 1
I0607 02:23:20.441444 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 02:23:20.441455 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 02:23:20.441468 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 02:23:20.441479 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 02:23:20.441491 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 02:23:20.441503 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 02:23:20.441514 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 02:23:20.441525 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 02:23:20.441537 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.954545
I0607 02:23:20.441550 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.952381
I0607 02:23:20.441563 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 0.483249 (* 0.3 = 0.144975 loss)
I0607 02:23:20.441577 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.157584 (* 0.3 = 0.0472752 loss)
I0607 02:23:20.441591 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.172179 (* 0.0272727 = 0.00469579 loss)
I0607 02:23:20.441606 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.3196 (* 0.0272727 = 0.00871635 loss)
I0607 02:23:20.441633 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.421596 (* 0.0272727 = 0.0114981 loss)
I0607 02:23:20.441648 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 0.571613 (* 0.0272727 = 0.0155894 loss)
I0607 02:23:20.441663 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.25483 (* 0.0272727 = 0.0342227 loss)
I0607 02:23:20.441676 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 0.549347 (* 0.0272727 = 0.0149822 loss)
I0607 02:23:20.441690 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.187699 (* 0.0272727 = 0.00511905 loss)
I0607 02:23:20.441704 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.00768127 (* 0.0272727 = 0.000209489 loss)
I0607 02:23:20.441718 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.00109497 (* 0.0272727 = 2.9863e-05 loss)
I0607 02:23:20.441730 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.000258299 (* 0.0272727 = 7.04453e-06 loss)
I0607 02:23:20.441740 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.000249946 (* 0.0272727 = 6.8167e-06 loss)
I0607 02:23:20.441750 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 5.45619e-05 (* 0.0272727 = 1.48805e-06 loss)
I0607 02:23:20.441758 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 3.74443e-05 (* 0.0272727 = 1.02121e-06 loss)
I0607 02:23:20.441773 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 1.56765e-05 (* 0.0272727 = 4.27541e-07 loss)
I0607 02:23:20.441787 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 9.04516e-06 (* 0.0272727 = 2.46686e-07 loss)
I0607 02:23:20.441804 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 1.37244e-05 (* 0.0272727 = 3.74301e-07 loss)
I0607 02:23:20.441818 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 1.09526e-05 (* 0.0272727 = 2.98708e-07 loss)
I0607 02:23:20.441833 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 8.53853e-06 (* 0.0272727 = 2.32869e-07 loss)
I0607 02:23:20.441859 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 1.13701e-05 (* 0.0272727 = 3.10093e-07 loss)
I0607 02:23:20.441884 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 7.10803e-06 (* 0.0272727 = 1.93855e-07 loss)
I0607 02:23:20.441900 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 9.79038e-06 (* 0.0272727 = 2.6701e-07 loss)
I0607 02:23:20.441915 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 6.97393e-06 (* 0.0272727 = 1.90198e-07 loss)
I0607 02:23:20.441926 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.952381
I0607 02:23:20.441939 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 02:23:20.441951 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 1
I0607 02:23:20.441962 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 1
I0607 02:23:20.441973 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 1
I0607 02:23:20.441985 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.75
I0607 02:23:20.441998 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 1
I0607 02:23:20.442019 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 1
I0607 02:23:20.442039 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 1
I0607 02:23:20.442051 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 1
I0607 02:23:20.442064 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 1
I0607 02:23:20.442075 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 1
I0607 02:23:20.442086 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 1
I0607 02:23:20.442098 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 1
I0607 02:23:20.442111 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 02:23:20.442121 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 02:23:20.442150 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 02:23:20.442170 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 02:23:20.442183 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 02:23:20.442194 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 02:23:20.442206 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 02:23:20.442217 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 02:23:20.442229 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 02:23:20.442240 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.988636
I0607 02:23:20.442252 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 1
I0607 02:23:20.442266 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.163739 (* 1 = 0.163739 loss)
I0607 02:23:20.442281 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.044667 (* 1 = 0.044667 loss)
I0607 02:23:20.442294 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.02927 (* 0.0909091 = 0.00266091 loss)
I0607 02:23:20.442309 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.115868 (* 0.0909091 = 0.0105334 loss)
I0607 02:23:20.442330 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.0135818 (* 0.0909091 = 0.00123471 loss)
I0607 02:23:20.442353 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.167037 (* 0.0909091 = 0.0151852 loss)
I0607 02:23:20.442368 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.716124 (* 0.0909091 = 0.0651022 loss)
I0607 02:23:20.442381 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.103933 (* 0.0909091 = 0.00944843 loss)
I0607 02:23:20.442395 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.0482807 (* 0.0909091 = 0.00438916 loss)
I0607 02:23:20.442410 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.00292155 (* 0.0909091 = 0.000265595 loss)
I0607 02:23:20.442425 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.00190807 (* 0.0909091 = 0.000173461 loss)
I0607 02:23:20.442438 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.000765758 (* 0.0909091 = 6.96143e-05 loss)
I0607 02:23:20.442452 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.000751371 (* 0.0909091 = 6.83064e-05 loss)
I0607 02:23:20.442466 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.00016395 (* 0.0909091 = 1.49046e-05 loss)
I0607 02:23:20.442481 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 8.67342e-05 (* 0.0909091 = 7.88493e-06 loss)
I0607 02:23:20.442495 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 4.39603e-05 (* 0.0909091 = 3.99639e-06 loss)
I0607 02:23:20.442505 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 2.98628e-05 (* 0.0909091 = 2.7148e-06 loss)
I0607 02:23:20.442515 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 2.29633e-05 (* 0.0909091 = 2.08757e-06 loss)
I0607 02:23:20.442533 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 4.528e-05 (* 0.0909091 = 4.11636e-06 loss)
I0607 02:23:20.442546 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 7.47728e-05 (* 0.0909091 = 6.79753e-06 loss)
I0607 02:23:20.442560 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 9.26778e-05 (* 0.0909091 = 8.42526e-06 loss)
I0607 02:23:20.442574 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 6.40627e-05 (* 0.0909091 = 5.82389e-06 loss)
I0607 02:23:20.442589 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 4.86436e-05 (* 0.0909091 = 4.42215e-06 loss)
I0607 02:23:20.442602 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 3.67122e-05 (* 0.0909091 = 3.33748e-06 loss)
I0607 02:23:20.442615 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.75
I0607 02:23:20.442637 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.75
I0607 02:23:20.442651 32403 solver.cpp:245] Train net output #149: total_confidence = 0.640258
I0607 02:23:20.442662 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.604143
I0607 02:23:20.442675 32403 sgd_solver.cpp:106] Iteration 6000, lr = 0.001
I0607 02:23:26.215728 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.2101 > 30) by scale factor 0.961228
I0607 02:23:33.199940 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.1456 > 30) by scale factor 0.87859
I0607 02:23:54.889017 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.5388 > 30) by scale factor 0.921976
I0607 02:23:57.209679 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 38.4399 > 30) by scale factor 0.78044
I0607 02:24:42.872828 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.4827 > 30) by scale factor 0.870001
I0607 02:25:25.445410 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.1179 > 30) by scale factor 0.934059
I0607 02:26:39.771256 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 59.7279 > 30) by scale factor 0.502278
I0607 02:28:01.040467 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 43.3254 > 30) by scale factor 0.692434
I0607 02:29:26.180598 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.6779 > 30) by scale factor 0.947033
I0607 02:29:47.515147 32403 solver.cpp:229] Iteration 6500, loss = 3.99983
I0607 02:29:47.515230 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.415094
I0607 02:29:47.515250 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.875
I0607 02:29:47.515264 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.625
I0607 02:29:47.515276 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.75
I0607 02:29:47.515290 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.5
I0607 02:29:47.515301 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.375
I0607 02:29:47.515313 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.375
I0607 02:29:47.515326 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.5
I0607 02:29:47.515338 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 1
I0607 02:29:47.515350 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.875
I0607 02:29:47.515362 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.875
I0607 02:29:47.515374 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.875
I0607 02:29:47.515388 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 02:29:47.515399 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.875
I0607 02:29:47.515411 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.875
I0607 02:29:47.515424 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 02:29:47.515436 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 02:29:47.515449 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 02:29:47.515460 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 02:29:47.515472 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 02:29:47.515485 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 02:29:47.515496 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 02:29:47.515508 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 02:29:47.515521 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.823864
I0607 02:29:47.515532 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.716981
I0607 02:29:47.515549 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.88057 (* 0.3 = 0.56417 loss)
I0607 02:29:47.515564 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.597796 (* 0.3 = 0.179339 loss)
I0607 02:29:47.515578 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 1.03074 (* 0.0272727 = 0.028111 loss)
I0607 02:29:47.515593 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.35653 (* 0.0272727 = 0.0369964 loss)
I0607 02:29:47.515606 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.96158 (* 0.0272727 = 0.0534975 loss)
I0607 02:29:47.515620 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.57605 (* 0.0272727 = 0.0429833 loss)
I0607 02:29:47.515635 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.75621 (* 0.0272727 = 0.0478967 loss)
I0607 02:29:47.515648 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.59144 (* 0.0272727 = 0.0434028 loss)
I0607 02:29:47.515662 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 2.33697 (* 0.0272727 = 0.0637356 loss)
I0607 02:29:47.515676 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.184522 (* 0.0272727 = 0.00503242 loss)
I0607 02:29:47.515691 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.290067 (* 0.0272727 = 0.00791092 loss)
I0607 02:29:47.515704 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.282071 (* 0.0272727 = 0.00769283 loss)
I0607 02:29:47.515718 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.265167 (* 0.0272727 = 0.00723184 loss)
I0607 02:29:47.515733 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.441084 (* 0.0272727 = 0.0120296 loss)
I0607 02:29:47.515794 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.35736 (* 0.0272727 = 0.00974617 loss)
I0607 02:29:47.515813 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.470831 (* 0.0272727 = 0.0128408 loss)
I0607 02:29:47.515830 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.00388608 (* 0.0272727 = 0.000105984 loss)
I0607 02:29:47.515851 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.00279086 (* 0.0272727 = 7.61144e-05 loss)
I0607 02:29:47.515864 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.0045525 (* 0.0272727 = 0.000124159 loss)
I0607 02:29:47.515878 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.00275768 (* 0.0272727 = 7.52095e-05 loss)
I0607 02:29:47.515893 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.00206888 (* 0.0272727 = 5.6424e-05 loss)
I0607 02:29:47.515911 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.00229969 (* 0.0272727 = 6.27189e-05 loss)
I0607 02:29:47.515925 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.0020152 (* 0.0272727 = 5.49601e-05 loss)
I0607 02:29:47.515940 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.00366098 (* 0.0272727 = 9.9845e-05 loss)
I0607 02:29:47.515952 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.622642
I0607 02:29:47.515965 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.75
I0607 02:29:47.515977 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.75
I0607 02:29:47.515990 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.875
I0607 02:29:47.516006 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.625
I0607 02:29:47.516018 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.5
I0607 02:29:47.516031 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.75
I0607 02:29:47.516052 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.5
I0607 02:29:47.516063 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.875
I0607 02:29:47.516075 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.875
I0607 02:29:47.516088 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 1
I0607 02:29:47.516106 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 1
I0607 02:29:47.516118 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.875
I0607 02:29:47.516130 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.875
I0607 02:29:47.516141 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.875
I0607 02:29:47.516154 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 02:29:47.516165 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 02:29:47.516177 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 02:29:47.516188 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 02:29:47.516201 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 02:29:47.516212 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 02:29:47.516224 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 02:29:47.516235 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 02:29:47.516247 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.880682
I0607 02:29:47.516259 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.754717
I0607 02:29:47.516273 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.26829 (* 0.3 = 0.380487 loss)
I0607 02:29:47.516288 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.42858 (* 0.3 = 0.128574 loss)
I0607 02:29:47.516301 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.821278 (* 0.0272727 = 0.0223985 loss)
I0607 02:29:47.516329 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.954549 (* 0.0272727 = 0.0260332 loss)
I0607 02:29:47.516351 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.961171 (* 0.0272727 = 0.0262138 loss)
I0607 02:29:47.516366 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.26004 (* 0.0272727 = 0.0343648 loss)
I0607 02:29:47.516379 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.72866 (* 0.0272727 = 0.0471452 loss)
I0607 02:29:47.516393 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.28694 (* 0.0272727 = 0.0350983 loss)
I0607 02:29:47.516417 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 1.51284 (* 0.0272727 = 0.0412592 loss)
I0607 02:29:47.516430 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.31743 (* 0.0272727 = 0.00865718 loss)
I0607 02:29:47.516444 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.263701 (* 0.0272727 = 0.00719184 loss)
I0607 02:29:47.516459 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.165655 (* 0.0272727 = 0.00451785 loss)
I0607 02:29:47.516474 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.254032 (* 0.0272727 = 0.00692814 loss)
I0607 02:29:47.516487 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.417332 (* 0.0272727 = 0.0113818 loss)
I0607 02:29:47.516497 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.390122 (* 0.0272727 = 0.0106397 loss)
I0607 02:29:47.516507 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.268256 (* 0.0272727 = 0.00731607 loss)
I0607 02:29:47.516518 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.0634395 (* 0.0272727 = 0.00173017 loss)
I0607 02:29:47.516532 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.0198373 (* 0.0272727 = 0.000541018 loss)
I0607 02:29:47.516546 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.00411156 (* 0.0272727 = 0.000112133 loss)
I0607 02:29:47.516561 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.00112216 (* 0.0272727 = 3.06043e-05 loss)
I0607 02:29:47.516574 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.0011575 (* 0.0272727 = 3.15681e-05 loss)
I0607 02:29:47.516589 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.00126146 (* 0.0272727 = 3.44034e-05 loss)
I0607 02:29:47.516603 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.00156649 (* 0.0272727 = 4.27225e-05 loss)
I0607 02:29:47.516618 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 0.000947229 (* 0.0272727 = 2.58335e-05 loss)
I0607 02:29:47.516629 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.849057
I0607 02:29:47.516641 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 0.875
I0607 02:29:47.516654 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.875
I0607 02:29:47.516665 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.875
I0607 02:29:47.516683 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.875
I0607 02:29:47.516695 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 1
I0607 02:29:47.516707 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.875
I0607 02:29:47.516719 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.75
I0607 02:29:47.516731 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 1
I0607 02:29:47.516746 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.875
I0607 02:29:47.516758 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 1
I0607 02:29:47.516770 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 1
I0607 02:29:47.516782 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.875
I0607 02:29:47.516793 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 1
I0607 02:29:47.516805 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 02:29:47.516818 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 02:29:47.516839 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 02:29:47.516854 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 02:29:47.516866 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 02:29:47.516878 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 02:29:47.516891 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 02:29:47.516901 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 02:29:47.516913 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 02:29:47.516926 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.948864
I0607 02:29:47.516937 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.90566
I0607 02:29:47.516952 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.628542 (* 1 = 0.628542 loss)
I0607 02:29:47.516965 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.233986 (* 1 = 0.233986 loss)
I0607 02:29:47.516979 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.413297 (* 0.0909091 = 0.0375725 loss)
I0607 02:29:47.516993 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.621038 (* 0.0909091 = 0.056458 loss)
I0607 02:29:47.517007 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.522676 (* 0.0909091 = 0.047516 loss)
I0607 02:29:47.517021 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.7048 (* 0.0909091 = 0.0640727 loss)
I0607 02:29:47.517035 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.389006 (* 0.0909091 = 0.0353641 loss)
I0607 02:29:47.517052 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.595513 (* 0.0909091 = 0.0541376 loss)
I0607 02:29:47.517067 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.516698 (* 0.0909091 = 0.0469725 loss)
I0607 02:29:47.517081 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.075119 (* 0.0909091 = 0.006829 loss)
I0607 02:29:47.517096 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.364598 (* 0.0909091 = 0.0331453 loss)
I0607 02:29:47.517109 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.141676 (* 0.0909091 = 0.0128796 loss)
I0607 02:29:47.517138 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.115002 (* 0.0909091 = 0.0104547 loss)
I0607 02:29:47.517153 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.169478 (* 0.0909091 = 0.0154071 loss)
I0607 02:29:47.517168 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.110199 (* 0.0909091 = 0.0100181 loss)
I0607 02:29:47.517181 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.0514715 (* 0.0909091 = 0.00467923 loss)
I0607 02:29:47.517195 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.0313631 (* 0.0909091 = 0.0028512 loss)
I0607 02:29:47.517210 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.0162998 (* 0.0909091 = 0.0014818 loss)
I0607 02:29:47.517225 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.00677546 (* 0.0909091 = 0.000615951 loss)
I0607 02:29:47.517238 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.00470883 (* 0.0909091 = 0.000428076 loss)
I0607 02:29:47.517252 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.00514078 (* 0.0909091 = 0.000467343 loss)
I0607 02:29:47.517273 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.00684691 (* 0.0909091 = 0.000622446 loss)
I0607 02:29:47.517283 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.00986525 (* 0.0909091 = 0.000896841 loss)
I0607 02:29:47.517299 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 0.00388226 (* 0.0909091 = 0.000352933 loss)
I0607 02:29:47.517313 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.625
I0607 02:29:47.517328 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.375
I0607 02:29:47.517351 32403 solver.cpp:245] Train net output #149: total_confidence = 0.319787
I0607 02:29:47.517364 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.186519
I0607 02:29:47.517379 32403 sgd_solver.cpp:106] Iteration 6500, lr = 0.001
I0607 02:29:53.289283 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 38.2201 > 30) by scale factor 0.784928
I0607 02:30:13.416713 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.2663 > 30) by scale factor 0.805017
I0607 02:30:56.120971 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.7832 > 30) by scale factor 0.794004
I0607 02:31:15.458703 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.3543 > 30) by scale factor 0.899433
I0607 02:31:20.097033 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 35.1919 > 30) by scale factor 0.85247
I0607 02:31:37.898475 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 44.2015 > 30) by scale factor 0.67871
I0607 02:32:17.364059 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 44.3498 > 30) by scale factor 0.67644
I0607 02:32:32.086036 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.6105 > 30) by scale factor 0.949051
I0607 02:34:06.487556 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 46.6784 > 30) by scale factor 0.642695
I0607 02:34:56.033797 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 35.6494 > 30) by scale factor 0.841529
I0607 02:36:05.853055 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.0634 > 30) by scale factor 0.809423
I0607 02:36:14.768156 32403 solver.cpp:229] Iteration 7000, loss = 4.2541
I0607 02:36:14.768225 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.571429
I0607 02:36:14.768244 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.625
I0607 02:36:14.768256 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.875
I0607 02:36:14.768270 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.875
I0607 02:36:14.768281 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.375
I0607 02:36:14.768295 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.5
I0607 02:36:14.768306 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.625
I0607 02:36:14.768318 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.75
I0607 02:36:14.768331 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.875
I0607 02:36:14.768342 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.875
I0607 02:36:14.768354 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.875
I0607 02:36:14.768367 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.875
I0607 02:36:14.768378 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 02:36:14.768391 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.875
I0607 02:36:14.768404 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.875
I0607 02:36:14.768416 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 02:36:14.768429 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 02:36:14.768440 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 02:36:14.768451 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 02:36:14.768463 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 02:36:14.768474 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 02:36:14.768486 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 02:36:14.768498 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 02:36:14.768510 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.840909
I0607 02:36:14.768522 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.767857
I0607 02:36:14.768538 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.38838 (* 0.3 = 0.416514 loss)
I0607 02:36:14.768558 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.510496 (* 0.3 = 0.153149 loss)
I0607 02:36:14.768573 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 1.40327 (* 0.0272727 = 0.038271 loss)
I0607 02:36:14.768587 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 0.957503 (* 0.0272727 = 0.0261137 loss)
I0607 02:36:14.768601 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 0.947623 (* 0.0272727 = 0.0258443 loss)
I0607 02:36:14.768615 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.84368 (* 0.0272727 = 0.0502823 loss)
I0607 02:36:14.768630 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.38447 (* 0.0272727 = 0.0377583 loss)
I0607 02:36:14.768643 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.59228 (* 0.0272727 = 0.0434258 loss)
I0607 02:36:14.768656 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 0.708892 (* 0.0272727 = 0.0193334 loss)
I0607 02:36:14.768671 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.613277 (* 0.0272727 = 0.0167257 loss)
I0607 02:36:14.768684 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.399215 (* 0.0272727 = 0.0108877 loss)
I0607 02:36:14.768699 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.49867 (* 0.0272727 = 0.0136001 loss)
I0607 02:36:14.768713 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.532328 (* 0.0272727 = 0.014518 loss)
I0607 02:36:14.768726 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.497388 (* 0.0272727 = 0.0135651 loss)
I0607 02:36:14.768785 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.682181 (* 0.0272727 = 0.0186049 loss)
I0607 02:36:14.768801 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.64818 (* 0.0272727 = 0.0176776 loss)
I0607 02:36:14.768816 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.00179795 (* 0.0272727 = 4.90351e-05 loss)
I0607 02:36:14.768831 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.000652016 (* 0.0272727 = 1.77823e-05 loss)
I0607 02:36:14.768844 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.000329199 (* 0.0272727 = 8.97815e-06 loss)
I0607 02:36:14.768858 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.000391568 (* 0.0272727 = 1.06791e-05 loss)
I0607 02:36:14.768872 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.000635902 (* 0.0272727 = 1.73428e-05 loss)
I0607 02:36:14.768887 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.000268344 (* 0.0272727 = 7.31846e-06 loss)
I0607 02:36:14.768900 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.000393599 (* 0.0272727 = 1.07345e-05 loss)
I0607 02:36:14.768914 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.00019534 (* 0.0272727 = 5.32746e-06 loss)
I0607 02:36:14.768928 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.660714
I0607 02:36:14.768939 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.5
I0607 02:36:14.768951 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.625
I0607 02:36:14.768964 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.75
I0607 02:36:14.768975 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.625
I0607 02:36:14.768987 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.75
I0607 02:36:14.768999 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.5
I0607 02:36:14.769011 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.875
I0607 02:36:14.769023 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.875
I0607 02:36:14.769035 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.875
I0607 02:36:14.769047 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.875
I0607 02:36:14.769059 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.875
I0607 02:36:14.769071 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.875
I0607 02:36:14.769083 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.875
I0607 02:36:14.769095 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.875
I0607 02:36:14.769107 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 02:36:14.769131 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 02:36:14.769145 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 02:36:14.769157 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 02:36:14.769170 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 02:36:14.769181 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 02:36:14.769192 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 02:36:14.769204 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 02:36:14.769217 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.869318
I0607 02:36:14.769228 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.875
I0607 02:36:14.769243 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.0967 (* 0.3 = 0.32901 loss)
I0607 02:36:14.769256 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.404565 (* 0.3 = 0.121369 loss)
I0607 02:36:14.769270 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 2.39535 (* 0.0272727 = 0.0653277 loss)
I0607 02:36:14.769297 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.891036 (* 0.0272727 = 0.024301 loss)
I0607 02:36:14.769312 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.570633 (* 0.0272727 = 0.0155627 loss)
I0607 02:36:14.769326 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.09359 (* 0.0272727 = 0.0298251 loss)
I0607 02:36:14.769340 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 0.871903 (* 0.0272727 = 0.0237792 loss)
I0607 02:36:14.769354 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.07315 (* 0.0272727 = 0.0292676 loss)
I0607 02:36:14.769368 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.396557 (* 0.0272727 = 0.0108152 loss)
I0607 02:36:14.769382 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.548292 (* 0.0272727 = 0.0149534 loss)
I0607 02:36:14.769397 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.157758 (* 0.0272727 = 0.0043025 loss)
I0607 02:36:14.769410 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.374431 (* 0.0272727 = 0.0102117 loss)
I0607 02:36:14.769424 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.335594 (* 0.0272727 = 0.00915256 loss)
I0607 02:36:14.769439 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.218635 (* 0.0272727 = 0.00596277 loss)
I0607 02:36:14.769454 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.31336 (* 0.0272727 = 0.00854618 loss)
I0607 02:36:14.769464 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.317377 (* 0.0272727 = 0.00865573 loss)
I0607 02:36:14.769474 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.0126345 (* 0.0272727 = 0.000344578 loss)
I0607 02:36:14.769490 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.00136807 (* 0.0272727 = 3.73111e-05 loss)
I0607 02:36:14.769503 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.000511633 (* 0.0272727 = 1.39536e-05 loss)
I0607 02:36:14.769517 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 4.76171e-05 (* 0.0272727 = 1.29865e-06 loss)
I0607 02:36:14.769531 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 2.02966e-05 (* 0.0272727 = 5.53545e-07 loss)
I0607 02:36:14.769546 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 7.00361e-06 (* 0.0272727 = 1.91007e-07 loss)
I0607 02:36:14.769559 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 3.03985e-06 (* 0.0272727 = 8.2905e-08 loss)
I0607 02:36:14.769573 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 1.34561e-05 (* 0.0272727 = 3.66985e-07 loss)
I0607 02:36:14.769585 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.910714
I0607 02:36:14.769603 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 0.75
I0607 02:36:14.769615 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.875
I0607 02:36:14.769628 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 1
I0607 02:36:14.769639 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.875
I0607 02:36:14.769651 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.875
I0607 02:36:14.769662 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 1
I0607 02:36:14.769675 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 1
I0607 02:36:14.769686 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.875
I0607 02:36:14.769698 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 1
I0607 02:36:14.769709 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.875
I0607 02:36:14.769721 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 1
I0607 02:36:14.769733 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 1
I0607 02:36:14.769744 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 1
I0607 02:36:14.769757 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 0.875
I0607 02:36:14.769778 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 02:36:14.769791 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 02:36:14.769805 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 02:36:14.769817 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 02:36:14.769829 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 02:36:14.769840 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 02:36:14.769851 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 02:36:14.769863 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 02:36:14.769876 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.965909
I0607 02:36:14.769887 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.964286
I0607 02:36:14.769901 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.542922 (* 1 = 0.542922 loss)
I0607 02:36:14.769915 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.190602 (* 1 = 0.190602 loss)
I0607 02:36:14.769929 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 1.24177 (* 0.0909091 = 0.112888 loss)
I0607 02:36:14.769943 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.249671 (* 0.0909091 = 0.0226974 loss)
I0607 02:36:14.769958 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.0715738 (* 0.0909091 = 0.00650671 loss)
I0607 02:36:14.769971 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.641107 (* 0.0909091 = 0.0582824 loss)
I0607 02:36:14.769985 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.312851 (* 0.0909091 = 0.028441 loss)
I0607 02:36:14.769999 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.213985 (* 0.0909091 = 0.0194532 loss)
I0607 02:36:14.770014 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.145242 (* 0.0909091 = 0.0132038 loss)
I0607 02:36:14.770027 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.439587 (* 0.0909091 = 0.0399624 loss)
I0607 02:36:14.770041 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.135793 (* 0.0909091 = 0.0123449 loss)
I0607 02:36:14.770056 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.18982 (* 0.0909091 = 0.0172564 loss)
I0607 02:36:14.770069 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.0611463 (* 0.0909091 = 0.00555876 loss)
I0607 02:36:14.770083 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.0854518 (* 0.0909091 = 0.00776835 loss)
I0607 02:36:14.770097 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.0564735 (* 0.0909091 = 0.00513395 loss)
I0607 02:36:14.770110 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.185265 (* 0.0909091 = 0.0168423 loss)
I0607 02:36:14.770125 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.0125778 (* 0.0909091 = 0.00114344 loss)
I0607 02:36:14.770139 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.00371255 (* 0.0909091 = 0.000337504 loss)
I0607 02:36:14.770153 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.00114065 (* 0.0909091 = 0.000103696 loss)
I0607 02:36:14.770167 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.000463313 (* 0.0909091 = 4.21194e-05 loss)
I0607 02:36:14.770181 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.000165427 (* 0.0909091 = 1.50388e-05 loss)
I0607 02:36:14.770195 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 7.81586e-05 (* 0.0909091 = 7.10533e-06 loss)
I0607 02:36:14.770210 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 1.86716e-05 (* 0.0909091 = 1.69742e-06 loss)
I0607 02:36:14.770223 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 1.75986e-05 (* 0.0909091 = 1.59987e-06 loss)
I0607 02:36:14.770236 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.625
I0607 02:36:14.770257 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.5
I0607 02:36:14.770270 32403 solver.cpp:245] Train net output #149: total_confidence = 0.412227
I0607 02:36:14.770282 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.334604
I0607 02:36:14.770295 32403 sgd_solver.cpp:106] Iteration 7000, lr = 0.001
I0607 02:36:24.438642 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.8328 > 30) by scale factor 0.97299
I0607 02:37:37.297462 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.5274 > 30) by scale factor 0.982723
I0607 02:38:05.164697 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 43.2778 > 30) by scale factor 0.693196
I0607 02:38:25.283568 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.8187 > 30) by scale factor 0.942841
I0607 02:39:07.891476 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 35.8632 > 30) by scale factor 0.836512
I0607 02:42:19.793535 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 36.5683 > 30) by scale factor 0.820383
I0607 02:42:41.855757 32403 solver.cpp:229] Iteration 7500, loss = 4.09854
I0607 02:42:41.855832 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.454545
I0607 02:42:41.855851 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.875
I0607 02:42:41.855865 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.875
I0607 02:42:41.855877 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.625
I0607 02:42:41.855890 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.5
I0607 02:42:41.855901 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.625
I0607 02:42:41.855913 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.375
I0607 02:42:41.855926 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.625
I0607 02:42:41.855938 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.75
I0607 02:42:41.855950 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.875
I0607 02:42:41.855963 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.875
I0607 02:42:41.855975 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.75
I0607 02:42:41.855988 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 02:42:41.856000 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 1
I0607 02:42:41.856012 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 1
I0607 02:42:41.856024 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 02:42:41.856036 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 02:42:41.856047 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 02:42:41.856060 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 02:42:41.856071 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 02:42:41.856082 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 02:42:41.856096 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 02:42:41.856106 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 02:42:41.856118 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.823864
I0607 02:42:41.856130 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.763636
I0607 02:42:41.856148 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.56992 (* 0.3 = 0.470976 loss)
I0607 02:42:41.856161 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.531248 (* 0.3 = 0.159374 loss)
I0607 02:42:41.856176 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.645446 (* 0.0272727 = 0.0176031 loss)
I0607 02:42:41.856190 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 0.504319 (* 0.0272727 = 0.0137542 loss)
I0607 02:42:41.856204 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.08828 (* 0.0272727 = 0.0296803 loss)
I0607 02:42:41.856218 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.38876 (* 0.0272727 = 0.0378754 loss)
I0607 02:42:41.856231 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.40574 (* 0.0272727 = 0.0383384 loss)
I0607 02:42:41.856245 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.5315 (* 0.0272727 = 0.0417681 loss)
I0607 02:42:41.856259 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 0.980691 (* 0.0272727 = 0.0267461 loss)
I0607 02:42:41.856273 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.973153 (* 0.0272727 = 0.0265405 loss)
I0607 02:42:41.856288 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.599924 (* 0.0272727 = 0.0163616 loss)
I0607 02:42:41.856302 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.582097 (* 0.0272727 = 0.0158754 loss)
I0607 02:42:41.856317 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.558811 (* 0.0272727 = 0.0152403 loss)
I0607 02:42:41.856330 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.402832 (* 0.0272727 = 0.0109863 loss)
I0607 02:42:41.856390 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.0338698 (* 0.0272727 = 0.000923723 loss)
I0607 02:42:41.856407 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.00765803 (* 0.0272727 = 0.000208855 loss)
I0607 02:42:41.856421 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.00186342 (* 0.0272727 = 5.08205e-05 loss)
I0607 02:42:41.856436 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.000723213 (* 0.0272727 = 1.9724e-05 loss)
I0607 02:42:41.856451 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.000330019 (* 0.0272727 = 9.00053e-06 loss)
I0607 02:42:41.856464 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.000217789 (* 0.0272727 = 5.93971e-06 loss)
I0607 02:42:41.856478 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.000140159 (* 0.0272727 = 3.82252e-06 loss)
I0607 02:42:41.856492 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 9.27976e-05 (* 0.0272727 = 2.53084e-06 loss)
I0607 02:42:41.856506 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 4.82472e-05 (* 0.0272727 = 1.31583e-06 loss)
I0607 02:42:41.856520 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 4.28383e-05 (* 0.0272727 = 1.16832e-06 loss)
I0607 02:42:41.856533 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.636364
I0607 02:42:41.856544 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 1
I0607 02:42:41.856555 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.875
I0607 02:42:41.856567 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.75
I0607 02:42:41.856580 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.375
I0607 02:42:41.856591 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.625
I0607 02:42:41.856603 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.5
I0607 02:42:41.856616 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.75
I0607 02:42:41.856626 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.75
I0607 02:42:41.856638 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.75
I0607 02:42:41.856650 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.875
I0607 02:42:41.856662 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.75
I0607 02:42:41.856673 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.875
I0607 02:42:41.856685 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 1
I0607 02:42:41.856698 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 1
I0607 02:42:41.856709 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 02:42:41.856720 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 02:42:41.856731 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 02:42:41.856744 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 02:42:41.856755 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 02:42:41.856766 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 02:42:41.856783 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 02:42:41.856796 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 02:42:41.856807 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.869318
I0607 02:42:41.856819 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.781818
I0607 02:42:41.856833 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.18395 (* 0.3 = 0.355186 loss)
I0607 02:42:41.856848 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.424432 (* 0.3 = 0.12733 loss)
I0607 02:42:41.856863 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.104314 (* 0.0272727 = 0.00284492 loss)
I0607 02:42:41.856889 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.196928 (* 0.0272727 = 0.00537075 loss)
I0607 02:42:41.856904 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.597211 (* 0.0272727 = 0.0162876 loss)
I0607 02:42:41.856919 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.36592 (* 0.0272727 = 0.0372523 loss)
I0607 02:42:41.856932 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.17657 (* 0.0272727 = 0.0320883 loss)
I0607 02:42:41.856945 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.34932 (* 0.0272727 = 0.0367997 loss)
I0607 02:42:41.856959 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.636254 (* 0.0272727 = 0.0173524 loss)
I0607 02:42:41.856973 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.959928 (* 0.0272727 = 0.0261798 loss)
I0607 02:42:41.856987 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.83615 (* 0.0272727 = 0.0228041 loss)
I0607 02:42:41.857002 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.453031 (* 0.0272727 = 0.0123554 loss)
I0607 02:42:41.857014 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.749085 (* 0.0272727 = 0.0204296 loss)
I0607 02:42:41.857028 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.515173 (* 0.0272727 = 0.0140502 loss)
I0607 02:42:41.857043 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.0483017 (* 0.0272727 = 0.00131732 loss)
I0607 02:42:41.857058 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.0111725 (* 0.0272727 = 0.000304706 loss)
I0607 02:42:41.857071 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.0025625 (* 0.0272727 = 6.98864e-05 loss)
I0607 02:42:41.857085 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.00113498 (* 0.0272727 = 3.09539e-05 loss)
I0607 02:42:41.857100 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.000858683 (* 0.0272727 = 2.34186e-05 loss)
I0607 02:42:41.857113 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.000673911 (* 0.0272727 = 1.83794e-05 loss)
I0607 02:42:41.857142 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.000250558 (* 0.0272727 = 6.83339e-06 loss)
I0607 02:42:41.857157 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 7.40989e-05 (* 0.0272727 = 2.02088e-06 loss)
I0607 02:42:41.857172 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 4.83029e-05 (* 0.0272727 = 1.31735e-06 loss)
I0607 02:42:41.857187 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 1.66754e-05 (* 0.0272727 = 4.54782e-07 loss)
I0607 02:42:41.857198 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.727273
I0607 02:42:41.857210 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 02:42:41.857221 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 1
I0607 02:42:41.857234 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.875
I0607 02:42:41.857245 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.75
I0607 02:42:41.857257 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.75
I0607 02:42:41.857269 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.75
I0607 02:42:41.857280 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.75
I0607 02:42:41.857292 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.875
I0607 02:42:41.857305 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.75
I0607 02:42:41.857316 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.75
I0607 02:42:41.857327 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.75
I0607 02:42:41.857338 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.75
I0607 02:42:41.857350 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 1
I0607 02:42:41.857362 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 02:42:41.857385 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 02:42:41.857398 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 02:42:41.857410 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 02:42:41.857421 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 02:42:41.857432 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 02:42:41.857445 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 02:42:41.857455 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 02:42:41.857467 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 02:42:41.857478 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.897727
I0607 02:42:41.857491 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.927273
I0607 02:42:41.857504 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.730182 (* 1 = 0.730182 loss)
I0607 02:42:41.857518 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.277888 (* 1 = 0.277888 loss)
I0607 02:42:41.857533 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.103631 (* 0.0909091 = 0.00942098 loss)
I0607 02:42:41.857547 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.150839 (* 0.0909091 = 0.0137126 loss)
I0607 02:42:41.857561 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.218493 (* 0.0909091 = 0.019863 loss)
I0607 02:42:41.857575 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.723736 (* 0.0909091 = 0.0657942 loss)
I0607 02:42:41.857589 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.880857 (* 0.0909091 = 0.0800779 loss)
I0607 02:42:41.857604 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 1.02242 (* 0.0909091 = 0.0929468 loss)
I0607 02:42:41.857614 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.524182 (* 0.0909091 = 0.0476529 loss)
I0607 02:42:41.857623 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.572466 (* 0.0909091 = 0.0520424 loss)
I0607 02:42:41.857637 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.884597 (* 0.0909091 = 0.080418 loss)
I0607 02:42:41.857651 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.461564 (* 0.0909091 = 0.0419604 loss)
I0607 02:42:41.857664 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.59671 (* 0.0909091 = 0.0542463 loss)
I0607 02:42:41.857678 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.551239 (* 0.0909091 = 0.0501126 loss)
I0607 02:42:41.857692 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.0888007 (* 0.0909091 = 0.00807279 loss)
I0607 02:42:41.857705 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.030731 (* 0.0909091 = 0.00279373 loss)
I0607 02:42:41.857720 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.0114653 (* 0.0909091 = 0.0010423 loss)
I0607 02:42:41.857734 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.00505202 (* 0.0909091 = 0.000459275 loss)
I0607 02:42:41.857748 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.00216296 (* 0.0909091 = 0.000196633 loss)
I0607 02:42:41.857761 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.000728519 (* 0.0909091 = 6.6229e-05 loss)
I0607 02:42:41.857775 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.0011544 (* 0.0909091 = 0.000104945 loss)
I0607 02:42:41.857789 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.000251507 (* 0.0909091 = 2.28643e-05 loss)
I0607 02:42:41.857802 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.000233034 (* 0.0909091 = 2.11849e-05 loss)
I0607 02:42:41.857815 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 8.88056e-05 (* 0.0909091 = 8.07323e-06 loss)
I0607 02:42:41.857831 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.625
I0607 02:42:41.857854 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.625
I0607 02:42:41.857867 32403 solver.cpp:245] Train net output #149: total_confidence = 0.442065
I0607 02:42:41.857879 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.452757
I0607 02:42:41.857892 32403 sgd_solver.cpp:106] Iteration 7500, lr = 0.001
I0607 02:42:54.631263 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.5259 > 30) by scale factor 0.799449
I0607 02:45:00.028216 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 47.1116 > 30) by scale factor 0.636786
I0607 02:45:10.101909 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 35.4963 > 30) by scale factor 0.845158
I0607 02:45:22.509477 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.1701 > 30) by scale factor 0.877961
I0607 02:45:39.562999 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 48.365 > 30) by scale factor 0.620284
I0607 02:47:45.086988 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 40.6637 > 30) by scale factor 0.737758
I0607 02:48:09.046766 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.3142 > 30) by scale factor 0.989635
I0607 02:48:33.037312 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 53.6338 > 30) by scale factor 0.559349
I0607 02:49:09.108983 32403 solver.cpp:229] Iteration 8000, loss = 4.04683
I0607 02:49:09.109109 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.413793
I0607 02:49:09.109132 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.75
I0607 02:49:09.109144 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.5
I0607 02:49:09.109158 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.375
I0607 02:49:09.109175 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.25
I0607 02:49:09.109200 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.25
I0607 02:49:09.109218 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.5
I0607 02:49:09.109231 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.625
I0607 02:49:09.109243 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.75
I0607 02:49:09.109256 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.75
I0607 02:49:09.109267 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.875
I0607 02:49:09.109279 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.875
I0607 02:49:09.109292 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 1
I0607 02:49:09.109304 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 1
I0607 02:49:09.109316 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.875
I0607 02:49:09.109328 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 0.875
I0607 02:49:09.109340 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 02:49:09.109352 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 02:49:09.109364 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 02:49:09.109376 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 02:49:09.109387 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 02:49:09.109400 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 02:49:09.109411 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 02:49:09.109423 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.795455
I0607 02:49:09.109436 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.758621
I0607 02:49:09.109452 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.89395 (* 0.3 = 0.568184 loss)
I0607 02:49:09.109465 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.675163 (* 0.3 = 0.202549 loss)
I0607 02:49:09.109480 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.959385 (* 0.0272727 = 0.026165 loss)
I0607 02:49:09.109494 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.59361 (* 0.0272727 = 0.043462 loss)
I0607 02:49:09.109508 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.99685 (* 0.0272727 = 0.0544596 loss)
I0607 02:49:09.109522 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 2.29037 (* 0.0272727 = 0.0624647 loss)
I0607 02:49:09.109536 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 2.39083 (* 0.0272727 = 0.0652044 loss)
I0607 02:49:09.109549 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 2.74462 (* 0.0272727 = 0.0748533 loss)
I0607 02:49:09.109563 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 0.946513 (* 0.0272727 = 0.025814 loss)
I0607 02:49:09.109577 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 1.0023 (* 0.0272727 = 0.0273353 loss)
I0607 02:49:09.109591 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.565696 (* 0.0272727 = 0.0154281 loss)
I0607 02:49:09.109606 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.206227 (* 0.0272727 = 0.00562438 loss)
I0607 02:49:09.109619 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.275548 (* 0.0272727 = 0.00751495 loss)
I0607 02:49:09.109633 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.0938891 (* 0.0272727 = 0.00256061 loss)
I0607 02:49:09.109671 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.110805 (* 0.0272727 = 0.00302196 loss)
I0607 02:49:09.109688 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.321563 (* 0.0272727 = 0.00876989 loss)
I0607 02:49:09.109701 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 1.0182 (* 0.0272727 = 0.027769 loss)
I0607 02:49:09.109716 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.0209419 (* 0.0272727 = 0.000571144 loss)
I0607 02:49:09.109730 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.00743059 (* 0.0272727 = 0.000202653 loss)
I0607 02:49:09.109745 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.00250238 (* 0.0272727 = 6.82467e-05 loss)
I0607 02:49:09.109760 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.00030289 (* 0.0272727 = 8.26062e-06 loss)
I0607 02:49:09.109773 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 7.92569e-05 (* 0.0272727 = 2.16155e-06 loss)
I0607 02:49:09.109787 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 3.3753e-05 (* 0.0272727 = 9.20535e-07 loss)
I0607 02:49:09.109802 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 2.86862e-05 (* 0.0272727 = 7.82352e-07 loss)
I0607 02:49:09.109814 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.672414
I0607 02:49:09.109827 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.875
I0607 02:49:09.109838 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.875
I0607 02:49:09.109850 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.75
I0607 02:49:09.109863 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.5
I0607 02:49:09.109876 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.5
I0607 02:49:09.109889 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.375
I0607 02:49:09.109901 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.75
I0607 02:49:09.109913 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.75
I0607 02:49:09.109925 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.875
I0607 02:49:09.109936 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 1
I0607 02:49:09.109948 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.875
I0607 02:49:09.109961 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 1
I0607 02:49:09.109972 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 1
I0607 02:49:09.109983 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.875
I0607 02:49:09.109995 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 0.875
I0607 02:49:09.110008 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 02:49:09.110018 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 02:49:09.110030 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 02:49:09.110043 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 02:49:09.110054 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 02:49:09.110065 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 02:49:09.110076 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 02:49:09.110088 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.880682
I0607 02:49:09.110100 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.896552
I0607 02:49:09.110113 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.17399 (* 0.3 = 0.352197 loss)
I0607 02:49:09.110131 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.429131 (* 0.3 = 0.128739 loss)
I0607 02:49:09.110146 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.29178 (* 0.0272727 = 0.00795763 loss)
I0607 02:49:09.110159 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 1.02101 (* 0.0272727 = 0.0278458 loss)
I0607 02:49:09.110184 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.788011 (* 0.0272727 = 0.0214912 loss)
I0607 02:49:09.110200 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.30172 (* 0.0272727 = 0.0355015 loss)
I0607 02:49:09.110214 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.82788 (* 0.0272727 = 0.0498512 loss)
I0607 02:49:09.110227 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 2.44853 (* 0.0272727 = 0.066778 loss)
I0607 02:49:09.110241 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.694323 (* 0.0272727 = 0.0189361 loss)
I0607 02:49:09.110255 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.616955 (* 0.0272727 = 0.0168261 loss)
I0607 02:49:09.110270 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.728249 (* 0.0272727 = 0.0198613 loss)
I0607 02:49:09.110283 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.08585 (* 0.0272727 = 0.00234136 loss)
I0607 02:49:09.110297 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.226146 (* 0.0272727 = 0.00616762 loss)
I0607 02:49:09.110311 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.0421054 (* 0.0272727 = 0.00114833 loss)
I0607 02:49:09.110326 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.139325 (* 0.0272727 = 0.00379977 loss)
I0607 02:49:09.110339 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.707645 (* 0.0272727 = 0.0192994 loss)
I0607 02:49:09.110353 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 1.05558 (* 0.0272727 = 0.0287886 loss)
I0607 02:49:09.110368 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.0173427 (* 0.0272727 = 0.000472982 loss)
I0607 02:49:09.110381 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.0139615 (* 0.0272727 = 0.000380768 loss)
I0607 02:49:09.110395 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.00377821 (* 0.0272727 = 0.000103042 loss)
I0607 02:49:09.110409 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.000138566 (* 0.0272727 = 3.77907e-06 loss)
I0607 02:49:09.110424 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.000200956 (* 0.0272727 = 5.48061e-06 loss)
I0607 02:49:09.110437 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 9.29842e-06 (* 0.0272727 = 2.53593e-07 loss)
I0607 02:49:09.110451 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 5.37935e-06 (* 0.0272727 = 1.4671e-07 loss)
I0607 02:49:09.110463 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.775862
I0607 02:49:09.110476 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 02:49:09.110486 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.875
I0607 02:49:09.110498 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.75
I0607 02:49:09.110510 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.875
I0607 02:49:09.110522 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.875
I0607 02:49:09.110533 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.75
I0607 02:49:09.110544 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.625
I0607 02:49:09.110556 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.75
I0607 02:49:09.110568 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.875
I0607 02:49:09.110579 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 1
I0607 02:49:09.110590 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.875
I0607 02:49:09.110602 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.875
I0607 02:49:09.110615 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 0.875
I0607 02:49:09.110625 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 0.875
I0607 02:49:09.110636 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 0.875
I0607 02:49:09.110658 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 02:49:09.110671 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 02:49:09.110683 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 02:49:09.110692 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 02:49:09.110698 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 02:49:09.110710 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 02:49:09.110723 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 02:49:09.110734 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.920455
I0607 02:49:09.110745 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.913793
I0607 02:49:09.110759 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.913511 (* 1 = 0.913511 loss)
I0607 02:49:09.110774 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.31699 (* 1 = 0.31699 loss)
I0607 02:49:09.110787 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.0788179 (* 0.0909091 = 0.00716527 loss)
I0607 02:49:09.110801 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.995976 (* 0.0909091 = 0.0905433 loss)
I0607 02:49:09.110816 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.499969 (* 0.0909091 = 0.0454518 loss)
I0607 02:49:09.110829 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.412614 (* 0.0909091 = 0.0375104 loss)
I0607 02:49:09.110843 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.494613 (* 0.0909091 = 0.0449648 loss)
I0607 02:49:09.110857 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 1.34538 (* 0.0909091 = 0.122308 loss)
I0607 02:49:09.110870 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.934568 (* 0.0909091 = 0.0849607 loss)
I0607 02:49:09.110884 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.66343 (* 0.0909091 = 0.0603118 loss)
I0607 02:49:09.110898 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.351529 (* 0.0909091 = 0.0319572 loss)
I0607 02:49:09.110913 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.0294772 (* 0.0909091 = 0.00267975 loss)
I0607 02:49:09.110929 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.179635 (* 0.0909091 = 0.0163305 loss)
I0607 02:49:09.110944 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.136766 (* 0.0909091 = 0.0124333 loss)
I0607 02:49:09.110957 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.250949 (* 0.0909091 = 0.0228136 loss)
I0607 02:49:09.110970 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.371058 (* 0.0909091 = 0.0337326 loss)
I0607 02:49:09.110985 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 1.3848 (* 0.0909091 = 0.125891 loss)
I0607 02:49:09.110998 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.00331665 (* 0.0909091 = 0.000301513 loss)
I0607 02:49:09.111011 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.00103846 (* 0.0909091 = 9.44055e-05 loss)
I0607 02:49:09.111026 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.000389805 (* 0.0909091 = 3.54369e-05 loss)
I0607 02:49:09.111039 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 7.96959e-05 (* 0.0909091 = 7.24509e-06 loss)
I0607 02:49:09.111053 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 5.17943e-05 (* 0.0909091 = 4.70858e-06 loss)
I0607 02:49:09.111066 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 2.00875e-05 (* 0.0909091 = 1.82614e-06 loss)
I0607 02:49:09.111080 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 2.26804e-05 (* 0.0909091 = 2.06186e-06 loss)
I0607 02:49:09.111093 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.375
I0607 02:49:09.111104 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.125
I0607 02:49:09.111125 32403 solver.cpp:245] Train net output #149: total_confidence = 0.317612
I0607 02:49:09.111138 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.324807
I0607 02:49:09.111151 32403 sgd_solver.cpp:106] Iteration 8000, lr = 0.001
I0607 02:50:04.456403 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.8499 > 30) by scale factor 0.886265
I0607 02:50:39.266543 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.9344 > 30) by scale factor 0.858753
I0607 02:53:30.344759 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.7264 > 30) by scale factor 0.795199
I0607 02:53:34.967525 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 47.9153 > 30) by scale factor 0.626105
I0607 02:54:12.867092 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 42.8741 > 30) by scale factor 0.699722
I0607 02:54:22.147125 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.6253 > 30) by scale factor 0.948607
I0607 02:54:35.289600 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 35.1525 > 30) by scale factor 0.853425
I0607 02:54:53.072298 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.954 > 30) by scale factor 0.91036
I0607 02:55:36.078258 32403 solver.cpp:229] Iteration 8500, loss = 4.08497
I0607 02:55:36.078400 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.452055
I0607 02:55:36.078421 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.625
I0607 02:55:36.078434 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.625
I0607 02:55:36.078447 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.5
I0607 02:55:36.078460 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.375
I0607 02:55:36.078472 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.25
I0607 02:55:36.078485 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.75
I0607 02:55:36.078497 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.625
I0607 02:55:36.078510 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.75
I0607 02:55:36.078521 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.875
I0607 02:55:36.078534 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.625
I0607 02:55:36.078547 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.875
I0607 02:55:36.078558 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.625
I0607 02:55:36.078572 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.625
I0607 02:55:36.078583 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.625
I0607 02:55:36.078595 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 0.875
I0607 02:55:36.078608 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 0.875
I0607 02:55:36.078620 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 0.875
I0607 02:55:36.078632 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 02:55:36.078644 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 02:55:36.078656 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 02:55:36.078668 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 02:55:36.078680 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 02:55:36.078693 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.761364
I0607 02:55:36.078706 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.753425
I0607 02:55:36.078722 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.64035 (* 0.3 = 0.492104 loss)
I0607 02:55:36.078737 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.733934 (* 0.3 = 0.22018 loss)
I0607 02:55:36.078752 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.903909 (* 0.0272727 = 0.0246521 loss)
I0607 02:55:36.078766 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.38087 (* 0.0272727 = 0.0376601 loss)
I0607 02:55:36.078780 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.14114 (* 0.0272727 = 0.0311221 loss)
I0607 02:55:36.078794 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.42265 (* 0.0272727 = 0.0387995 loss)
I0607 02:55:36.078809 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.8172 (* 0.0272727 = 0.0495599 loss)
I0607 02:55:36.078822 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.04556 (* 0.0272727 = 0.0285152 loss)
I0607 02:55:36.078836 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 0.918881 (* 0.0272727 = 0.0250604 loss)
I0607 02:55:36.078850 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 1.28806 (* 0.0272727 = 0.035129 loss)
I0607 02:55:36.078865 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.505791 (* 0.0272727 = 0.0137943 loss)
I0607 02:55:36.078881 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.935394 (* 0.0272727 = 0.0255108 loss)
I0607 02:55:36.078896 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.678609 (* 0.0272727 = 0.0185075 loss)
I0607 02:55:36.078909 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 1.09164 (* 0.0272727 = 0.029772 loss)
I0607 02:55:36.078951 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 1.12305 (* 0.0272727 = 0.0306287 loss)
I0607 02:55:36.078966 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.957076 (* 0.0272727 = 0.0261021 loss)
I0607 02:55:36.078980 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.966493 (* 0.0272727 = 0.0263589 loss)
I0607 02:55:36.078994 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.436502 (* 0.0272727 = 0.0119046 loss)
I0607 02:55:36.079008 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.905422 (* 0.0272727 = 0.0246933 loss)
I0607 02:55:36.079023 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.00113688 (* 0.0272727 = 3.10057e-05 loss)
I0607 02:55:36.079038 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.0021779 (* 0.0272727 = 5.93973e-05 loss)
I0607 02:55:36.079052 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.000386963 (* 0.0272727 = 1.05535e-05 loss)
I0607 02:55:36.079067 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.00125624 (* 0.0272727 = 3.42611e-05 loss)
I0607 02:55:36.079082 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.000543297 (* 0.0272727 = 1.48172e-05 loss)
I0607 02:55:36.079093 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.506849
I0607 02:55:36.079107 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.75
I0607 02:55:36.079119 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.875
I0607 02:55:36.079131 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.75
I0607 02:55:36.079144 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.5
I0607 02:55:36.079155 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.5
I0607 02:55:36.079167 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.625
I0607 02:55:36.079179 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.875
I0607 02:55:36.079190 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.75
I0607 02:55:36.079202 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.875
I0607 02:55:36.079215 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.75
I0607 02:55:36.079226 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.625
I0607 02:55:36.079238 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.625
I0607 02:55:36.079251 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.75
I0607 02:55:36.079262 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.75
I0607 02:55:36.079274 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 0.75
I0607 02:55:36.079287 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 0.875
I0607 02:55:36.079298 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 0.875
I0607 02:55:36.079310 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 02:55:36.079321 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 02:55:36.079334 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 02:55:36.079344 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 02:55:36.079356 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 02:55:36.079368 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.778409
I0607 02:55:36.079380 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.835616
I0607 02:55:36.079394 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.30259 (* 0.3 = 0.390778 loss)
I0607 02:55:36.079407 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.59191 (* 0.3 = 0.177573 loss)
I0607 02:55:36.079426 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.613122 (* 0.0272727 = 0.0167215 loss)
I0607 02:55:36.079440 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.835402 (* 0.0272727 = 0.0227837 loss)
I0607 02:55:36.079465 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.701869 (* 0.0272727 = 0.0191419 loss)
I0607 02:55:36.079481 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.16219 (* 0.0272727 = 0.0316962 loss)
I0607 02:55:36.079494 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.38219 (* 0.0272727 = 0.037696 loss)
I0607 02:55:36.079509 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 0.949457 (* 0.0272727 = 0.0258943 loss)
I0607 02:55:36.079522 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.759091 (* 0.0272727 = 0.0207025 loss)
I0607 02:55:36.079536 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 1.07727 (* 0.0272727 = 0.02938 loss)
I0607 02:55:36.079550 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.495031 (* 0.0272727 = 0.0135008 loss)
I0607 02:55:36.079563 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.887153 (* 0.0272727 = 0.0241951 loss)
I0607 02:55:36.079577 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.636606 (* 0.0272727 = 0.017362 loss)
I0607 02:55:36.079591 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.851347 (* 0.0272727 = 0.0232186 loss)
I0607 02:55:36.079604 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.958687 (* 0.0272727 = 0.026146 loss)
I0607 02:55:36.079618 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.997096 (* 0.0272727 = 0.0271935 loss)
I0607 02:55:36.079632 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.831936 (* 0.0272727 = 0.0226892 loss)
I0607 02:55:36.079646 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.519229 (* 0.0272727 = 0.0141608 loss)
I0607 02:55:36.079659 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.522484 (* 0.0272727 = 0.0142496 loss)
I0607 02:55:36.079674 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.0120594 (* 0.0272727 = 0.000328894 loss)
I0607 02:55:36.079689 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.00286813 (* 0.0272727 = 7.82217e-05 loss)
I0607 02:55:36.079702 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.00178423 (* 0.0272727 = 4.86609e-05 loss)
I0607 02:55:36.079718 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.00237579 (* 0.0272727 = 6.47943e-05 loss)
I0607 02:55:36.079728 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 0.000785264 (* 0.0272727 = 2.14163e-05 loss)
I0607 02:55:36.079741 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.739726
I0607 02:55:36.079753 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 0.875
I0607 02:55:36.079766 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.875
I0607 02:55:36.079777 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.875
I0607 02:55:36.079789 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.875
I0607 02:55:36.079802 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.875
I0607 02:55:36.079813 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.875
I0607 02:55:36.079825 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 1
I0607 02:55:36.079836 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.875
I0607 02:55:36.079849 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.875
I0607 02:55:36.079860 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.75
I0607 02:55:36.079872 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.875
I0607 02:55:36.079884 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.75
I0607 02:55:36.079895 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 0.625
I0607 02:55:36.079907 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 0.625
I0607 02:55:36.079918 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 0.875
I0607 02:55:36.079943 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 0.75
I0607 02:55:36.079957 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 0.875
I0607 02:55:36.079969 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 02:55:36.079982 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 02:55:36.079993 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 02:55:36.080004 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 02:55:36.080016 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 02:55:36.080029 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.880682
I0607 02:55:36.080040 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.890411
I0607 02:55:36.080054 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.840553 (* 1 = 0.840553 loss)
I0607 02:55:36.080068 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.388605 (* 1 = 0.388605 loss)
I0607 02:55:36.080082 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.287301 (* 0.0909091 = 0.0261183 loss)
I0607 02:55:36.080096 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.566 (* 0.0909091 = 0.0514545 loss)
I0607 02:55:36.080111 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.207069 (* 0.0909091 = 0.0188245 loss)
I0607 02:55:36.080123 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.269275 (* 0.0909091 = 0.0244795 loss)
I0607 02:55:36.080137 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.702617 (* 0.0909091 = 0.0638742 loss)
I0607 02:55:36.080152 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.533897 (* 0.0909091 = 0.0485361 loss)
I0607 02:55:36.080165 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.219773 (* 0.0909091 = 0.0199794 loss)
I0607 02:55:36.080178 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.422429 (* 0.0909091 = 0.0384026 loss)
I0607 02:55:36.080193 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.277654 (* 0.0909091 = 0.0252413 loss)
I0607 02:55:36.080206 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.656826 (* 0.0909091 = 0.0597114 loss)
I0607 02:55:36.080219 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.623732 (* 0.0909091 = 0.0567029 loss)
I0607 02:55:36.080234 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.640803 (* 0.0909091 = 0.0582548 loss)
I0607 02:55:36.080247 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.829787 (* 0.0909091 = 0.0754352 loss)
I0607 02:55:36.080261 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.947391 (* 0.0909091 = 0.0861265 loss)
I0607 02:55:36.080274 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.612164 (* 0.0909091 = 0.0556513 loss)
I0607 02:55:36.080288 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.624642 (* 0.0909091 = 0.0567856 loss)
I0607 02:55:36.080302 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.755932 (* 0.0909091 = 0.0687211 loss)
I0607 02:55:36.080317 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.00295 (* 0.0909091 = 0.000268182 loss)
I0607 02:55:36.080330 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.00137821 (* 0.0909091 = 0.000125292 loss)
I0607 02:55:36.080344 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.000665562 (* 0.0909091 = 6.05057e-05 loss)
I0607 02:55:36.080358 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.000252947 (* 0.0909091 = 2.29951e-05 loss)
I0607 02:55:36.080373 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 0.000112654 (* 0.0909091 = 1.02413e-05 loss)
I0607 02:55:36.080384 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.5
I0607 02:55:36.080396 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.625
I0607 02:55:36.080417 32403 solver.cpp:245] Train net output #149: total_confidence = 0.440856
I0607 02:55:36.080427 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.409168
I0607 02:55:36.080436 32403 sgd_solver.cpp:106] Iteration 8500, lr = 0.001
I0607 02:56:33.675796 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.8168 > 30) by scale factor 0.914165
I0607 02:57:51.072860 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.2479 > 30) by scale factor 0.930293
I0607 03:00:47.649933 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.0914 > 30) by scale factor 0.90658
I0607 03:00:56.960692 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 55.0733 > 30) by scale factor 0.544729
I0607 03:01:15.516362 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.9915 > 30) by scale factor 0.909326
I0607 03:01:50.357636 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.809 > 30) by scale factor 0.793462
I0607 03:02:03.156123 32403 solver.cpp:229] Iteration 9000, loss = 4.02535
I0607 03:02:03.156198 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.388889
I0607 03:02:03.156216 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.375
I0607 03:02:03.156229 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.625
I0607 03:02:03.156242 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.375
I0607 03:02:03.156255 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.25
I0607 03:02:03.156266 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.5
I0607 03:02:03.156280 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.625
I0607 03:02:03.156291 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.5
I0607 03:02:03.156303 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.75
I0607 03:02:03.156316 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.75
I0607 03:02:03.156327 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.875
I0607 03:02:03.156340 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.875
I0607 03:02:03.156352 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 03:02:03.156364 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 1
I0607 03:02:03.156376 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 1
I0607 03:02:03.156388 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 03:02:03.156400 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 03:02:03.156412 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 03:02:03.156424 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 03:02:03.156436 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 03:02:03.156450 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 03:02:03.156462 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 03:02:03.156474 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 03:02:03.156486 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.806818
I0607 03:02:03.156498 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.722222
I0607 03:02:03.156514 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.88647 (* 0.3 = 0.565941 loss)
I0607 03:02:03.156529 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.616186 (* 0.3 = 0.184856 loss)
I0607 03:02:03.156544 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 2.02357 (* 0.0272727 = 0.0551882 loss)
I0607 03:02:03.156558 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.41536 (* 0.0272727 = 0.0386008 loss)
I0607 03:02:03.156572 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.69308 (* 0.0272727 = 0.0461748 loss)
I0607 03:02:03.156586 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 2.12888 (* 0.0272727 = 0.0580602 loss)
I0607 03:02:03.156600 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.90054 (* 0.0272727 = 0.0518328 loss)
I0607 03:02:03.156615 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.59029 (* 0.0272727 = 0.0433716 loss)
I0607 03:02:03.156628 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 1.84511 (* 0.0272727 = 0.0503211 loss)
I0607 03:02:03.156642 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.75508 (* 0.0272727 = 0.0205931 loss)
I0607 03:02:03.156656 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.736099 (* 0.0272727 = 0.0200754 loss)
I0607 03:02:03.156671 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.402175 (* 0.0272727 = 0.0109684 loss)
I0607 03:02:03.156684 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.510176 (* 0.0272727 = 0.0139139 loss)
I0607 03:02:03.156698 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.5186 (* 0.0272727 = 0.0141436 loss)
I0607 03:02:03.156759 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.0817173 (* 0.0272727 = 0.00222865 loss)
I0607 03:02:03.156776 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.0448581 (* 0.0272727 = 0.0012234 loss)
I0607 03:02:03.156791 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.025443 (* 0.0272727 = 0.0006939 loss)
I0607 03:02:03.156805 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.0185494 (* 0.0272727 = 0.000505892 loss)
I0607 03:02:03.156821 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.0145342 (* 0.0272727 = 0.000396387 loss)
I0607 03:02:03.156834 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.00613389 (* 0.0272727 = 0.000167288 loss)
I0607 03:02:03.156848 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.00821544 (* 0.0272727 = 0.000224058 loss)
I0607 03:02:03.156862 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.00376731 (* 0.0272727 = 0.000102745 loss)
I0607 03:02:03.156877 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.00819613 (* 0.0272727 = 0.000223531 loss)
I0607 03:02:03.156890 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.00690822 (* 0.0272727 = 0.000188406 loss)
I0607 03:02:03.156903 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.666667
I0607 03:02:03.156916 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.75
I0607 03:02:03.156929 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.875
I0607 03:02:03.156940 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.75
I0607 03:02:03.156952 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.75
I0607 03:02:03.156965 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.625
I0607 03:02:03.156976 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.875
I0607 03:02:03.156988 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.625
I0607 03:02:03.157001 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.875
I0607 03:02:03.157012 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 1
I0607 03:02:03.157023 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 1
I0607 03:02:03.157035 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.875
I0607 03:02:03.157047 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.875
I0607 03:02:03.157059 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 1
I0607 03:02:03.157071 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 1
I0607 03:02:03.157083 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 03:02:03.157094 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 03:02:03.157106 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 03:02:03.157130 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 03:02:03.157145 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 03:02:03.157157 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 03:02:03.157169 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 03:02:03.157184 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 03:02:03.157196 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.886364
I0607 03:02:03.157208 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.888889
I0607 03:02:03.157223 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.19473 (* 0.3 = 0.358419 loss)
I0607 03:02:03.157238 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.399406 (* 0.3 = 0.119822 loss)
I0607 03:02:03.157251 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.680762 (* 0.0272727 = 0.0185662 loss)
I0607 03:02:03.157266 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.487878 (* 0.0272727 = 0.0133058 loss)
I0607 03:02:03.157294 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.671612 (* 0.0272727 = 0.0183167 loss)
I0607 03:02:03.157308 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 0.846206 (* 0.0272727 = 0.0230783 loss)
I0607 03:02:03.157322 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.18529 (* 0.0272727 = 0.032326 loss)
I0607 03:02:03.157337 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 0.744246 (* 0.0272727 = 0.0202976 loss)
I0607 03:02:03.157351 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 1.91828 (* 0.0272727 = 0.0523168 loss)
I0607 03:02:03.157366 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.918433 (* 0.0272727 = 0.0250482 loss)
I0607 03:02:03.157379 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.28786 (* 0.0272727 = 0.00785074 loss)
I0607 03:02:03.157393 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.240886 (* 0.0272727 = 0.00656961 loss)
I0607 03:02:03.157407 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.421663 (* 0.0272727 = 0.0114999 loss)
I0607 03:02:03.157421 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.369137 (* 0.0272727 = 0.0100674 loss)
I0607 03:02:03.157435 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.0349603 (* 0.0272727 = 0.000953463 loss)
I0607 03:02:03.157449 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.004815 (* 0.0272727 = 0.000131318 loss)
I0607 03:02:03.157464 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.00228426 (* 0.0272727 = 6.22979e-05 loss)
I0607 03:02:03.157479 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.000306582 (* 0.0272727 = 8.36133e-06 loss)
I0607 03:02:03.157493 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.00010351 (* 0.0272727 = 2.823e-06 loss)
I0607 03:02:03.157507 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 5.63587e-05 (* 0.0272727 = 1.53706e-06 loss)
I0607 03:02:03.157518 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 2.13402e-05 (* 0.0272727 = 5.82004e-07 loss)
I0607 03:02:03.157528 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 7.15229e-05 (* 0.0272727 = 1.95063e-06 loss)
I0607 03:02:03.157543 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 1.10273e-05 (* 0.0272727 = 3.00745e-07 loss)
I0607 03:02:03.157558 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 1.9671e-05 (* 0.0272727 = 5.36482e-07 loss)
I0607 03:02:03.157570 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.833333
I0607 03:02:03.157583 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 0.875
I0607 03:02:03.157595 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 1
I0607 03:02:03.157608 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 1
I0607 03:02:03.157619 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 1
I0607 03:02:03.157631 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.875
I0607 03:02:03.157642 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.875
I0607 03:02:03.157655 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.75
I0607 03:02:03.157667 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.875
I0607 03:02:03.157680 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.875
I0607 03:02:03.157691 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.875
I0607 03:02:03.157703 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.875
I0607 03:02:03.157714 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.875
I0607 03:02:03.157727 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 1
I0607 03:02:03.157738 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 03:02:03.157750 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 03:02:03.157773 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 03:02:03.157786 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 03:02:03.157799 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 03:02:03.157812 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 03:02:03.157824 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 03:02:03.157836 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 03:02:03.157848 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 03:02:03.157860 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.948864
I0607 03:02:03.157872 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.944444
I0607 03:02:03.157886 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.559285 (* 1 = 0.559285 loss)
I0607 03:02:03.157899 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.184691 (* 1 = 0.184691 loss)
I0607 03:02:03.157914 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.303509 (* 0.0909091 = 0.0275917 loss)
I0607 03:02:03.157928 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.0634062 (* 0.0909091 = 0.0057642 loss)
I0607 03:02:03.157943 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.114116 (* 0.0909091 = 0.0103742 loss)
I0607 03:02:03.157956 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.112933 (* 0.0909091 = 0.0102666 loss)
I0607 03:02:03.157970 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.354838 (* 0.0909091 = 0.032258 loss)
I0607 03:02:03.157984 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.627171 (* 0.0909091 = 0.0570156 loss)
I0607 03:02:03.157999 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.908939 (* 0.0909091 = 0.0826308 loss)
I0607 03:02:03.158012 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.448173 (* 0.0909091 = 0.040743 loss)
I0607 03:02:03.158026 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.266745 (* 0.0909091 = 0.0242495 loss)
I0607 03:02:03.158041 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.373739 (* 0.0909091 = 0.0339763 loss)
I0607 03:02:03.158054 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.541791 (* 0.0909091 = 0.0492537 loss)
I0607 03:02:03.158068 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.662855 (* 0.0909091 = 0.0602596 loss)
I0607 03:02:03.158082 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.0024614 (* 0.0909091 = 0.000223763 loss)
I0607 03:02:03.158097 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.00255118 (* 0.0909091 = 0.000231926 loss)
I0607 03:02:03.158110 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.000288129 (* 0.0909091 = 2.61936e-05 loss)
I0607 03:02:03.158125 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.000205374 (* 0.0909091 = 1.86703e-05 loss)
I0607 03:02:03.158139 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.000174205 (* 0.0909091 = 1.58368e-05 loss)
I0607 03:02:03.158154 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.000132076 (* 0.0909091 = 1.20069e-05 loss)
I0607 03:02:03.158169 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 9.2927e-05 (* 0.0909091 = 8.44791e-06 loss)
I0607 03:02:03.158182 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 7.86464e-05 (* 0.0909091 = 7.14967e-06 loss)
I0607 03:02:03.158196 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 3.55258e-05 (* 0.0909091 = 3.22962e-06 loss)
I0607 03:02:03.158211 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 2.32913e-05 (* 0.0909091 = 2.11739e-06 loss)
I0607 03:02:03.158223 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.75
I0607 03:02:03.158251 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.5
I0607 03:02:03.158264 32403 solver.cpp:245] Train net output #149: total_confidence = 0.32867
I0607 03:02:03.158277 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.307927
I0607 03:02:03.158290 32403 sgd_solver.cpp:106] Iteration 9000, lr = 0.001
I0607 03:03:30.959283 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.5506 > 30) by scale factor 0.798921
I0607 03:04:03.464707 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.078 > 30) by scale factor 0.997406
I0607 03:05:37.811100 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.7466 > 30) by scale factor 0.975719
I0607 03:05:40.130204 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.6805 > 30) by scale factor 0.86504
I0607 03:06:18.859380 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.4107 > 30) by scale factor 0.955088
I0607 03:08:30.074731 32403 solver.cpp:229] Iteration 9500, loss = 4.15728
I0607 03:08:30.074894 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.46875
I0607 03:08:30.074915 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.75
I0607 03:08:30.074929 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.875
I0607 03:08:30.074941 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.625
I0607 03:08:30.074954 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.5
I0607 03:08:30.074966 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.5
I0607 03:08:30.074978 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.625
I0607 03:08:30.074990 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.5
I0607 03:08:30.075002 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.625
I0607 03:08:30.075014 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.75
I0607 03:08:30.075027 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.75
I0607 03:08:30.075040 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.75
I0607 03:08:30.075052 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 03:08:30.075064 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.75
I0607 03:08:30.075076 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.75
I0607 03:08:30.075088 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 0.875
I0607 03:08:30.075100 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 03:08:30.075114 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 03:08:30.075124 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 03:08:30.075137 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 03:08:30.075148 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 03:08:30.075160 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 03:08:30.075172 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 03:08:30.075184 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.801136
I0607 03:08:30.075196 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.671875
I0607 03:08:30.075213 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.73141 (* 0.3 = 0.519424 loss)
I0607 03:08:30.075227 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.687142 (* 0.3 = 0.206143 loss)
I0607 03:08:30.075242 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.762213 (* 0.0272727 = 0.0207876 loss)
I0607 03:08:30.075258 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 0.821376 (* 0.0272727 = 0.0224012 loss)
I0607 03:08:30.075271 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.76744 (* 0.0272727 = 0.0482029 loss)
I0607 03:08:30.075285 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.40058 (* 0.0272727 = 0.0381977 loss)
I0607 03:08:30.075299 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.86144 (* 0.0272727 = 0.0507665 loss)
I0607 03:08:30.075314 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.2916 (* 0.0272727 = 0.0352254 loss)
I0607 03:08:30.075327 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 1.35475 (* 0.0272727 = 0.0369477 loss)
I0607 03:08:30.075341 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 1.08671 (* 0.0272727 = 0.0296376 loss)
I0607 03:08:30.075356 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.688112 (* 0.0272727 = 0.0187667 loss)
I0607 03:08:30.075369 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.758753 (* 0.0272727 = 0.0206933 loss)
I0607 03:08:30.075384 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.760008 (* 0.0272727 = 0.0207275 loss)
I0607 03:08:30.075398 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.622638 (* 0.0272727 = 0.016981 loss)
I0607 03:08:30.075434 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 1.10664 (* 0.0272727 = 0.0301811 loss)
I0607 03:08:30.075449 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.826864 (* 0.0272727 = 0.0225508 loss)
I0607 03:08:30.075464 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.313463 (* 0.0272727 = 0.008549 loss)
I0607 03:08:30.075479 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.0684303 (* 0.0272727 = 0.00186628 loss)
I0607 03:08:30.075494 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.0296671 (* 0.0272727 = 0.000809102 loss)
I0607 03:08:30.075507 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.00998153 (* 0.0272727 = 0.000272224 loss)
I0607 03:08:30.075522 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.0032113 (* 0.0272727 = 8.7581e-05 loss)
I0607 03:08:30.075536 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.00180268 (* 0.0272727 = 4.91641e-05 loss)
I0607 03:08:30.075551 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.000500985 (* 0.0272727 = 1.36632e-05 loss)
I0607 03:08:30.075565 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.000127944 (* 0.0272727 = 3.48937e-06 loss)
I0607 03:08:30.075577 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.59375
I0607 03:08:30.075590 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.875
I0607 03:08:30.075603 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 1
I0607 03:08:30.075615 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.625
I0607 03:08:30.075628 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.5
I0607 03:08:30.075639 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.625
I0607 03:08:30.075650 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.75
I0607 03:08:30.075662 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.5
I0607 03:08:30.075675 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.75
I0607 03:08:30.075686 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.75
I0607 03:08:30.075698 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 1
I0607 03:08:30.075711 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.75
I0607 03:08:30.075722 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.75
I0607 03:08:30.075734 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.875
I0607 03:08:30.075747 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.75
I0607 03:08:30.075758 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 0.875
I0607 03:08:30.075770 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 03:08:30.075783 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 03:08:30.075793 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 03:08:30.075805 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 03:08:30.075816 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 03:08:30.075829 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 03:08:30.075839 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 03:08:30.075851 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.840909
I0607 03:08:30.075863 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.828125
I0607 03:08:30.075877 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.2568 (* 0.3 = 0.377041 loss)
I0607 03:08:30.075893 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.502016 (* 0.3 = 0.150605 loss)
I0607 03:08:30.075907 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.58894 (* 0.0272727 = 0.016062 loss)
I0607 03:08:30.075922 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.0846568 (* 0.0272727 = 0.00230882 loss)
I0607 03:08:30.075948 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 1.234 (* 0.0272727 = 0.0336545 loss)
I0607 03:08:30.075963 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.20435 (* 0.0272727 = 0.032846 loss)
I0607 03:08:30.075978 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 0.757126 (* 0.0272727 = 0.0206489 loss)
I0607 03:08:30.075991 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 0.958097 (* 0.0272727 = 0.0261299 loss)
I0607 03:08:30.076005 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 1.2922 (* 0.0272727 = 0.0352419 loss)
I0607 03:08:30.076020 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.851595 (* 0.0272727 = 0.0232253 loss)
I0607 03:08:30.076032 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.523217 (* 0.0272727 = 0.0142695 loss)
I0607 03:08:30.076046 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.399432 (* 0.0272727 = 0.0108936 loss)
I0607 03:08:30.076061 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.817057 (* 0.0272727 = 0.0222834 loss)
I0607 03:08:30.076074 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.850821 (* 0.0272727 = 0.0232042 loss)
I0607 03:08:30.076087 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.652036 (* 0.0272727 = 0.0177828 loss)
I0607 03:08:30.076102 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.538166 (* 0.0272727 = 0.0146773 loss)
I0607 03:08:30.076115 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.379128 (* 0.0272727 = 0.0103398 loss)
I0607 03:08:30.076130 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.139722 (* 0.0272727 = 0.00381061 loss)
I0607 03:08:30.076144 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.0425268 (* 0.0272727 = 0.00115982 loss)
I0607 03:08:30.076159 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.0152844 (* 0.0272727 = 0.000416846 loss)
I0607 03:08:30.076172 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.00180236 (* 0.0272727 = 4.91552e-05 loss)
I0607 03:08:30.076187 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.00280581 (* 0.0272727 = 7.65221e-05 loss)
I0607 03:08:30.076201 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.000249684 (* 0.0272727 = 6.80957e-06 loss)
I0607 03:08:30.076215 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 0.000157776 (* 0.0272727 = 4.30298e-06 loss)
I0607 03:08:30.076227 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.859375
I0607 03:08:30.076239 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 0.875
I0607 03:08:30.076251 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 1
I0607 03:08:30.076263 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.875
I0607 03:08:30.076275 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.875
I0607 03:08:30.076287 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 1
I0607 03:08:30.076298 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.875
I0607 03:08:30.076310 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 1
I0607 03:08:30.076323 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.875
I0607 03:08:30.076333 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 1
I0607 03:08:30.076345 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.875
I0607 03:08:30.076357 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.75
I0607 03:08:30.076370 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 1
I0607 03:08:30.076380 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 0.875
I0607 03:08:30.076392 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 0.75
I0607 03:08:30.076405 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 0.875
I0607 03:08:30.076416 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 03:08:30.076437 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 03:08:30.076452 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 03:08:30.076462 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 03:08:30.076474 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 03:08:30.076486 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 03:08:30.076498 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 03:08:30.076509 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.948864
I0607 03:08:30.076522 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.9375
I0607 03:08:30.076535 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.559262 (* 1 = 0.559262 loss)
I0607 03:08:30.076550 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.213693 (* 1 = 0.213693 loss)
I0607 03:08:30.076565 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.279751 (* 0.0909091 = 0.0254319 loss)
I0607 03:08:30.076580 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.0594431 (* 0.0909091 = 0.00540391 loss)
I0607 03:08:30.076593 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.341015 (* 0.0909091 = 0.0310013 loss)
I0607 03:08:30.076607 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 1.05694 (* 0.0909091 = 0.0960857 loss)
I0607 03:08:30.076622 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.0870042 (* 0.0909091 = 0.00790947 loss)
I0607 03:08:30.076638 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.19532 (* 0.0909091 = 0.0177564 loss)
I0607 03:08:30.076653 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.275766 (* 0.0909091 = 0.0250696 loss)
I0607 03:08:30.076666 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.425491 (* 0.0909091 = 0.038681 loss)
I0607 03:08:30.076680 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.171394 (* 0.0909091 = 0.0155813 loss)
I0607 03:08:30.076694 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.685734 (* 0.0909091 = 0.0623395 loss)
I0607 03:08:30.076709 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.874587 (* 0.0909091 = 0.0795079 loss)
I0607 03:08:30.076722 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.206236 (* 0.0909091 = 0.0187487 loss)
I0607 03:08:30.076736 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.268447 (* 0.0909091 = 0.0244043 loss)
I0607 03:08:30.076750 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.570155 (* 0.0909091 = 0.0518323 loss)
I0607 03:08:30.076764 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.284835 (* 0.0909091 = 0.0258941 loss)
I0607 03:08:30.076778 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.0769175 (* 0.0909091 = 0.0069925 loss)
I0607 03:08:30.076792 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.024014 (* 0.0909091 = 0.00218309 loss)
I0607 03:08:30.076807 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.00975449 (* 0.0909091 = 0.000886772 loss)
I0607 03:08:30.076820 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.00287122 (* 0.0909091 = 0.00026102 loss)
I0607 03:08:30.076834 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.00102555 (* 0.0909091 = 9.32322e-05 loss)
I0607 03:08:30.076848 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.00100208 (* 0.0909091 = 9.10984e-05 loss)
I0607 03:08:30.076861 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 0.000256985 (* 0.0909091 = 2.33622e-05 loss)
I0607 03:08:30.076874 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.625
I0607 03:08:30.076885 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.5
I0607 03:08:30.076896 32403 solver.cpp:245] Train net output #149: total_confidence = 0.405042
I0607 03:08:30.076917 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.409715
I0607 03:08:30.076936 32403 sgd_solver.cpp:106] Iteration 9500, lr = 0.001
I0607 03:08:59.873636 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.0704 > 30) by scale factor 0.935441
I0607 03:10:52.131314 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 55.9122 > 30) by scale factor 0.536555
I0607 03:11:09.131134 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.1503 > 30) by scale factor 0.995015
I0607 03:11:23.027222 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 44.4929 > 30) by scale factor 0.674265
I0607 03:12:58.873034 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 40.5749 > 30) by scale factor 0.739373
I0607 03:14:56.448321 32403 solver.cpp:456] Snapshotting to binary proto file /mnt/snapshots/mixed_lstm21_iter_10000.caffemodel
I0607 03:14:57.149739 32403 sgd_solver.cpp:273] Snapshotting solver state to binary proto file /mnt/snapshots/mixed_lstm21_iter_10000.solverstate
I0607 03:14:57.432561 32403 solver.cpp:338] Iteration 10000, Testing net (#0)
I0607 03:15:56.218528 32403 solver.cpp:393] Test loss: 2.66748
I0607 03:15:56.218659 32403 solver.cpp:406] Test net output #0: loss1/accuracy = 0.630131
I0607 03:15:56.218680 32403 solver.cpp:406] Test net output #1: loss1/accuracy01 = 0.789
I0607 03:15:56.218694 32403 solver.cpp:406] Test net output #2: loss1/accuracy02 = 0.66
I0607 03:15:56.218708 32403 solver.cpp:406] Test net output #3: loss1/accuracy03 = 0.549
I0607 03:15:56.218720 32403 solver.cpp:406] Test net output #4: loss1/accuracy04 = 0.479
I0607 03:15:56.218734 32403 solver.cpp:406] Test net output #5: loss1/accuracy05 = 0.531
I0607 03:15:56.218745 32403 solver.cpp:406] Test net output #6: loss1/accuracy06 = 0.709
I0607 03:15:56.218758 32403 solver.cpp:406] Test net output #7: loss1/accuracy07 = 0.859
I0607 03:15:56.218770 32403 solver.cpp:406] Test net output #8: loss1/accuracy08 = 0.924
I0607 03:15:56.218783 32403 solver.cpp:406] Test net output #9: loss1/accuracy09 = 0.968
I0607 03:15:56.218796 32403 solver.cpp:406] Test net output #10: loss1/accuracy10 = 0.984
I0607 03:15:56.218808 32403 solver.cpp:406] Test net output #11: loss1/accuracy11 = 0.993
I0607 03:15:56.218821 32403 solver.cpp:406] Test net output #12: loss1/accuracy12 = 0.998
I0607 03:15:56.218832 32403 solver.cpp:406] Test net output #13: loss1/accuracy13 = 0.999
I0607 03:15:56.218844 32403 solver.cpp:406] Test net output #14: loss1/accuracy14 = 1
I0607 03:15:56.218857 32403 solver.cpp:406] Test net output #15: loss1/accuracy15 = 1
I0607 03:15:56.218868 32403 solver.cpp:406] Test net output #16: loss1/accuracy16 = 1
I0607 03:15:56.218883 32403 solver.cpp:406] Test net output #17: loss1/accuracy17 = 1
I0607 03:15:56.218894 32403 solver.cpp:406] Test net output #18: loss1/accuracy18 = 1
I0607 03:15:56.218906 32403 solver.cpp:406] Test net output #19: loss1/accuracy19 = 1
I0607 03:15:56.218917 32403 solver.cpp:406] Test net output #20: loss1/accuracy20 = 1
I0607 03:15:56.218930 32403 solver.cpp:406] Test net output #21: loss1/accuracy21 = 1
I0607 03:15:56.218940 32403 solver.cpp:406] Test net output #22: loss1/accuracy22 = 1
I0607 03:15:56.218952 32403 solver.cpp:406] Test net output #23: loss1/accuracy_incl_empty = 0.891775
I0607 03:15:56.218964 32403 solver.cpp:406] Test net output #24: loss1/accuracy_top3 = 0.845198
I0607 03:15:56.218981 32403 solver.cpp:406] Test net output #25: loss1/cross_entropy_loss = 1.35502 (* 0.3 = 0.406506 loss)
I0607 03:15:56.218997 32403 solver.cpp:406] Test net output #26: loss1/cross_entropy_loss_incl_empty = 0.396125 (* 0.3 = 0.118837 loss)
I0607 03:15:56.219010 32403 solver.cpp:406] Test net output #27: loss1/loss01 = 0.907314 (* 0.0272727 = 0.0247449 loss)
I0607 03:15:56.219024 32403 solver.cpp:406] Test net output #28: loss1/loss02 = 1.35505 (* 0.0272727 = 0.036956 loss)
I0607 03:15:56.219038 32403 solver.cpp:406] Test net output #29: loss1/loss03 = 1.64857 (* 0.0272727 = 0.0449611 loss)
I0607 03:15:56.219053 32403 solver.cpp:406] Test net output #30: loss1/loss04 = 1.79718 (* 0.0272727 = 0.0490139 loss)
I0607 03:15:56.219066 32403 solver.cpp:406] Test net output #31: loss1/loss05 = 1.57312 (* 0.0272727 = 0.0429032 loss)
I0607 03:15:56.219084 32403 solver.cpp:406] Test net output #32: loss1/loss06 = 0.999086 (* 0.0272727 = 0.0272478 loss)
I0607 03:15:56.219097 32403 solver.cpp:406] Test net output #33: loss1/loss07 = 0.514741 (* 0.0272727 = 0.0140384 loss)
I0607 03:15:56.219112 32403 solver.cpp:406] Test net output #34: loss1/loss08 = 0.268164 (* 0.0272727 = 0.00731357 loss)
I0607 03:15:56.219126 32403 solver.cpp:406] Test net output #35: loss1/loss09 = 0.164613 (* 0.0272727 = 0.00448946 loss)
I0607 03:15:56.219141 32403 solver.cpp:406] Test net output #36: loss1/loss10 = 0.07694 (* 0.0272727 = 0.00209836 loss)
I0607 03:15:56.219154 32403 solver.cpp:406] Test net output #37: loss1/loss11 = 0.0295525 (* 0.0272727 = 0.000805978 loss)
I0607 03:15:56.219168 32403 solver.cpp:406] Test net output #38: loss1/loss12 = 0.0171311 (* 0.0272727 = 0.000467211 loss)
I0607 03:15:56.219182 32403 solver.cpp:406] Test net output #39: loss1/loss13 = 0.0113658 (* 0.0272727 = 0.000309977 loss)
I0607 03:15:56.219226 32403 solver.cpp:406] Test net output #40: loss1/loss14 = 0.0084773 (* 0.0272727 = 0.000231199 loss)
I0607 03:15:56.219243 32403 solver.cpp:406] Test net output #41: loss1/loss15 = 0.00595444 (* 0.0272727 = 0.000162394 loss)
I0607 03:15:56.219256 32403 solver.cpp:406] Test net output #42: loss1/loss16 = 0.00467185 (* 0.0272727 = 0.000127414 loss)
I0607 03:15:56.219280 32403 solver.cpp:406] Test net output #43: loss1/loss17 = 0.00428843 (* 0.0272727 = 0.000116957 loss)
I0607 03:15:56.219293 32403 solver.cpp:406] Test net output #44: loss1/loss18 = 0.00372191 (* 0.0272727 = 0.000101507 loss)
I0607 03:15:56.219307 32403 solver.cpp:406] Test net output #45: loss1/loss19 = 0.0029988 (* 0.0272727 = 8.17853e-05 loss)
I0607 03:15:56.219321 32403 solver.cpp:406] Test net output #46: loss1/loss20 = 0.00272479 (* 0.0272727 = 7.43123e-05 loss)
I0607 03:15:56.219336 32403 solver.cpp:406] Test net output #47: loss1/loss21 = 0.00241344 (* 0.0272727 = 6.5821e-05 loss)
I0607 03:15:56.219349 32403 solver.cpp:406] Test net output #48: loss1/loss22 = 0.00219156 (* 0.0272727 = 5.97697e-05 loss)
I0607 03:15:56.219362 32403 solver.cpp:406] Test net output #49: loss2/accuracy = 0.770348
I0607 03:15:56.219374 32403 solver.cpp:406] Test net output #50: loss2/accuracy01 = 0.87
I0607 03:15:56.219385 32403 solver.cpp:406] Test net output #51: loss2/accuracy02 = 0.827
I0607 03:15:56.219398 32403 solver.cpp:406] Test net output #52: loss2/accuracy03 = 0.759
I0607 03:15:56.219409 32403 solver.cpp:406] Test net output #53: loss2/accuracy04 = 0.647
I0607 03:15:56.219420 32403 solver.cpp:406] Test net output #54: loss2/accuracy05 = 0.66
I0607 03:15:56.219432 32403 solver.cpp:406] Test net output #55: loss2/accuracy06 = 0.793
I0607 03:15:56.219444 32403 solver.cpp:406] Test net output #56: loss2/accuracy07 = 0.887
I0607 03:15:56.219455 32403 solver.cpp:406] Test net output #57: loss2/accuracy08 = 0.936
I0607 03:15:56.219467 32403 solver.cpp:406] Test net output #58: loss2/accuracy09 = 0.967
I0607 03:15:56.219478 32403 solver.cpp:406] Test net output #59: loss2/accuracy10 = 0.984
I0607 03:15:56.219491 32403 solver.cpp:406] Test net output #60: loss2/accuracy11 = 0.995
I0607 03:15:56.219501 32403 solver.cpp:406] Test net output #61: loss2/accuracy12 = 0.999
I0607 03:15:56.219513 32403 solver.cpp:406] Test net output #62: loss2/accuracy13 = 1
I0607 03:15:56.219524 32403 solver.cpp:406] Test net output #63: loss2/accuracy14 = 1
I0607 03:15:56.219535 32403 solver.cpp:406] Test net output #64: loss2/accuracy15 = 1
I0607 03:15:56.219547 32403 solver.cpp:406] Test net output #65: loss2/accuracy16 = 1
I0607 03:15:56.219558 32403 solver.cpp:406] Test net output #66: loss2/accuracy17 = 1
I0607 03:15:56.219569 32403 solver.cpp:406] Test net output #67: loss2/accuracy18 = 1
I0607 03:15:56.219580 32403 solver.cpp:406] Test net output #68: loss2/accuracy19 = 1
I0607 03:15:56.219593 32403 solver.cpp:406] Test net output #69: loss2/accuracy20 = 1
I0607 03:15:56.219604 32403 solver.cpp:406] Test net output #70: loss2/accuracy21 = 1
I0607 03:15:56.219614 32403 solver.cpp:406] Test net output #71: loss2/accuracy22 = 1
I0607 03:15:56.219626 32403 solver.cpp:406] Test net output #72: loss2/accuracy_incl_empty = 0.931137
I0607 03:15:56.219637 32403 solver.cpp:406] Test net output #73: loss2/accuracy_top3 = 0.913854
I0607 03:15:56.219651 32403 solver.cpp:406] Test net output #74: loss2/cross_entropy_loss = 0.914585 (* 0.3 = 0.274375 loss)
I0607 03:15:56.219666 32403 solver.cpp:406] Test net output #75: loss2/cross_entropy_loss_incl_empty = 0.269764 (* 0.3 = 0.0809293 loss)
I0607 03:15:56.219679 32403 solver.cpp:406] Test net output #76: loss2/loss01 = 0.581944 (* 0.0272727 = 0.0158712 loss)
I0607 03:15:56.219693 32403 solver.cpp:406] Test net output #77: loss2/loss02 = 0.829859 (* 0.0272727 = 0.0226325 loss)
I0607 03:15:56.219718 32403 solver.cpp:406] Test net output #78: loss2/loss03 = 0.96869 (* 0.0272727 = 0.0264188 loss)
I0607 03:15:56.219733 32403 solver.cpp:406] Test net output #79: loss2/loss04 = 1.22586 (* 0.0272727 = 0.0334325 loss)
I0607 03:15:56.219746 32403 solver.cpp:406] Test net output #80: loss2/loss05 = 1.11979 (* 0.0272727 = 0.0305396 loss)
I0607 03:15:56.219760 32403 solver.cpp:406] Test net output #81: loss2/loss06 = 0.743364 (* 0.0272727 = 0.0202736 loss)
I0607 03:15:56.219774 32403 solver.cpp:406] Test net output #82: loss2/loss07 = 0.41306 (* 0.0272727 = 0.0112653 loss)
I0607 03:15:56.219789 32403 solver.cpp:406] Test net output #83: loss2/loss08 = 0.224855 (* 0.0272727 = 0.00613242 loss)
I0607 03:15:56.219801 32403 solver.cpp:406] Test net output #84: loss2/loss09 = 0.144538 (* 0.0272727 = 0.00394196 loss)
I0607 03:15:56.219815 32403 solver.cpp:406] Test net output #85: loss2/loss10 = 0.0689792 (* 0.0272727 = 0.00188125 loss)
I0607 03:15:56.219830 32403 solver.cpp:406] Test net output #86: loss2/loss11 = 0.0204512 (* 0.0272727 = 0.000557759 loss)
I0607 03:15:56.219843 32403 solver.cpp:406] Test net output #87: loss2/loss12 = 0.00905175 (* 0.0272727 = 0.000246866 loss)
I0607 03:15:56.219857 32403 solver.cpp:406] Test net output #88: loss2/loss13 = 0.00551515 (* 0.0272727 = 0.000150413 loss)
I0607 03:15:56.219871 32403 solver.cpp:406] Test net output #89: loss2/loss14 = 0.00350481 (* 0.0272727 = 9.55858e-05 loss)
I0607 03:15:56.219882 32403 solver.cpp:406] Test net output #90: loss2/loss15 = 0.00226541 (* 0.0272727 = 6.17838e-05 loss)
I0607 03:15:56.219892 32403 solver.cpp:406] Test net output #91: loss2/loss16 = 0.00172728 (* 0.0272727 = 4.71077e-05 loss)
I0607 03:15:56.219905 32403 solver.cpp:406] Test net output #92: loss2/loss17 = 0.00148299 (* 0.0272727 = 4.04452e-05 loss)
I0607 03:15:56.219919 32403 solver.cpp:406] Test net output #93: loss2/loss18 = 0.00132669 (* 0.0272727 = 3.61826e-05 loss)
I0607 03:15:56.219936 32403 solver.cpp:406] Test net output #94: loss2/loss19 = 0.000889047 (* 0.0272727 = 2.42467e-05 loss)
I0607 03:15:56.219950 32403 solver.cpp:406] Test net output #95: loss2/loss20 = 0.00098839 (* 0.0272727 = 2.69561e-05 loss)
I0607 03:15:56.219965 32403 solver.cpp:406] Test net output #96: loss2/loss21 = 0.000907412 (* 0.0272727 = 2.47476e-05 loss)
I0607 03:15:56.219979 32403 solver.cpp:406] Test net output #97: loss2/loss22 = 0.000863701 (* 0.0272727 = 2.35555e-05 loss)
I0607 03:15:56.219991 32403 solver.cpp:406] Test net output #98: loss3/accuracy = 0.843876
I0607 03:15:56.220010 32403 solver.cpp:406] Test net output #99: loss3/accuracy01 = 0.892
I0607 03:15:56.220021 32403 solver.cpp:406] Test net output #100: loss3/accuracy02 = 0.858
I0607 03:15:56.220032 32403 solver.cpp:406] Test net output #101: loss3/accuracy03 = 0.864
I0607 03:15:56.220044 32403 solver.cpp:406] Test net output #102: loss3/accuracy04 = 0.845
I0607 03:15:56.220055 32403 solver.cpp:406] Test net output #103: loss3/accuracy05 = 0.841
I0607 03:15:56.220072 32403 solver.cpp:406] Test net output #104: loss3/accuracy06 = 0.861
I0607 03:15:56.220083 32403 solver.cpp:406] Test net output #105: loss3/accuracy07 = 0.9
I0607 03:15:56.220095 32403 solver.cpp:406] Test net output #106: loss3/accuracy08 = 0.945
I0607 03:15:56.220106 32403 solver.cpp:406] Test net output #107: loss3/accuracy09 = 0.966
I0607 03:15:56.220118 32403 solver.cpp:406] Test net output #108: loss3/accuracy10 = 0.981
I0607 03:15:56.220134 32403 solver.cpp:406] Test net output #109: loss3/accuracy11 = 0.992
I0607 03:15:56.220145 32403 solver.cpp:406] Test net output #110: loss3/accuracy12 = 0.997
I0607 03:15:56.220156 32403 solver.cpp:406] Test net output #111: loss3/accuracy13 = 0.999
I0607 03:15:56.220168 32403 solver.cpp:406] Test net output #112: loss3/accuracy14 = 0.998
I0607 03:15:56.220180 32403 solver.cpp:406] Test net output #113: loss3/accuracy15 = 1
I0607 03:15:56.220191 32403 solver.cpp:406] Test net output #114: loss3/accuracy16 = 1
I0607 03:15:56.220212 32403 solver.cpp:406] Test net output #115: loss3/accuracy17 = 1
I0607 03:15:56.220226 32403 solver.cpp:406] Test net output #116: loss3/accuracy18 = 1
I0607 03:15:56.220237 32403 solver.cpp:406] Test net output #117: loss3/accuracy19 = 1
I0607 03:15:56.220248 32403 solver.cpp:406] Test net output #118: loss3/accuracy20 = 1
I0607 03:15:56.220260 32403 solver.cpp:406] Test net output #119: loss3/accuracy21 = 1
I0607 03:15:56.220271 32403 solver.cpp:406] Test net output #120: loss3/accuracy22 = 1
I0607 03:15:56.220283 32403 solver.cpp:406] Test net output #121: loss3/accuracy_incl_empty = 0.949364
I0607 03:15:56.220294 32403 solver.cpp:406] Test net output #122: loss3/accuracy_top3 = 0.92665
I0607 03:15:56.220307 32403 solver.cpp:406] Test net output #123: loss3/cross_entropy_loss = 0.698785 (* 1 = 0.698785 loss)
I0607 03:15:56.220321 32403 solver.cpp:406] Test net output #124: loss3/cross_entropy_loss_incl_empty = 0.228596 (* 1 = 0.228596 loss)
I0607 03:15:56.220335 32403 solver.cpp:406] Test net output #125: loss3/loss01 = 0.505858 (* 0.0909091 = 0.0459871 loss)
I0607 03:15:56.220348 32403 solver.cpp:406] Test net output #126: loss3/loss02 = 0.660579 (* 0.0909091 = 0.0600526 loss)
I0607 03:15:56.220362 32403 solver.cpp:406] Test net output #127: loss3/loss03 = 0.678707 (* 0.0909091 = 0.0617007 loss)
I0607 03:15:56.220376 32403 solver.cpp:406] Test net output #128: loss3/loss04 = 0.734272 (* 0.0909091 = 0.066752 loss)
I0607 03:15:56.220389 32403 solver.cpp:406] Test net output #129: loss3/loss05 = 0.720842 (* 0.0909091 = 0.0655311 loss)
I0607 03:15:56.220403 32403 solver.cpp:406] Test net output #130: loss3/loss06 = 0.559921 (* 0.0909091 = 0.0509019 loss)
I0607 03:15:56.220417 32403 solver.cpp:406] Test net output #131: loss3/loss07 = 0.386703 (* 0.0909091 = 0.0351548 loss)
I0607 03:15:56.220432 32403 solver.cpp:406] Test net output #132: loss3/loss08 = 0.209953 (* 0.0909091 = 0.0190867 loss)
I0607 03:15:56.220445 32403 solver.cpp:406] Test net output #133: loss3/loss09 = 0.128653 (* 0.0909091 = 0.0116957 loss)
I0607 03:15:56.220459 32403 solver.cpp:406] Test net output #134: loss3/loss10 = 0.0798922 (* 0.0909091 = 0.00726293 loss)
I0607 03:15:56.220474 32403 solver.cpp:406] Test net output #135: loss3/loss11 = 0.0276749 (* 0.0909091 = 0.0025159 loss)
I0607 03:15:56.220487 32403 solver.cpp:406] Test net output #136: loss3/loss12 = 0.0139577 (* 0.0909091 = 0.00126888 loss)
I0607 03:15:56.220501 32403 solver.cpp:406] Test net output #137: loss3/loss13 = 0.00655282 (* 0.0909091 = 0.000595711 loss)
I0607 03:15:56.220515 32403 solver.cpp:406] Test net output #138: loss3/loss14 = 0.00417791 (* 0.0909091 = 0.00037981 loss)
I0607 03:15:56.220530 32403 solver.cpp:406] Test net output #139: loss3/loss15 = 0.00138452 (* 0.0909091 = 0.000125866 loss)
I0607 03:15:56.220543 32403 solver.cpp:406] Test net output #140: loss3/loss16 = 0.000884413 (* 0.0909091 = 8.04012e-05 loss)
I0607 03:15:56.220557 32403 solver.cpp:406] Test net output #141: loss3/loss17 = 0.000564858 (* 0.0909091 = 5.13507e-05 loss)
I0607 03:15:56.220571 32403 solver.cpp:406] Test net output #142: loss3/loss18 = 0.000477967 (* 0.0909091 = 4.34516e-05 loss)
I0607 03:15:56.220585 32403 solver.cpp:406] Test net output #143: loss3/loss19 = 0.00044601 (* 0.0909091 = 4.05464e-05 loss)
I0607 03:15:56.220599 32403 solver.cpp:406] Test net output #144: loss3/loss20 = 0.000473218 (* 0.0909091 = 4.30199e-05 loss)
I0607 03:15:56.220613 32403 solver.cpp:406] Test net output #145: loss3/loss21 = 0.000500766 (* 0.0909091 = 4.55242e-05 loss)
I0607 03:15:56.220628 32403 solver.cpp:406] Test net output #146: loss3/loss22 = 0.00042475 (* 0.0909091 = 3.86136e-05 loss)
I0607 03:15:56.220639 32403 solver.cpp:406] Test net output #147: total_accuracy = 0.573
I0607 03:15:56.220649 32403 solver.cpp:406] Test net output #148: total_accuracy_not_rec = 0.575
I0607 03:15:56.220664 32403 solver.cpp:406] Test net output #149: total_confidence = 0.485268
I0607 03:15:56.220695 32403 solver.cpp:406] Test net output #150: total_confidence_not_rec = 0.470175
I0607 03:15:56.220708 32403 solver.cpp:338] Iteration 10000, Testing net (#1)
I0607 03:16:54.825963 32403 solver.cpp:393] Test loss: 3.7363
I0607 03:16:54.826119 32403 solver.cpp:406] Test net output #0: loss1/accuracy = 0.568212
I0607 03:16:54.826141 32403 solver.cpp:406] Test net output #1: loss1/accuracy01 = 0.772
I0607 03:16:54.826155 32403 solver.cpp:406] Test net output #2: loss1/accuracy02 = 0.651
I0607 03:16:54.826167 32403 solver.cpp:406] Test net output #3: loss1/accuracy03 = 0.521
I0607 03:16:54.826180 32403 solver.cpp:406] Test net output #4: loss1/accuracy04 = 0.427
I0607 03:16:54.826194 32403 solver.cpp:406] Test net output #5: loss1/accuracy05 = 0.478
I0607 03:16:54.826205 32403 solver.cpp:406] Test net output #6: loss1/accuracy06 = 0.637
I0607 03:16:54.826218 32403 solver.cpp:406] Test net output #7: loss1/accuracy07 = 0.759
I0607 03:16:54.826231 32403 solver.cpp:406] Test net output #8: loss1/accuracy08 = 0.797
I0607 03:16:54.826243 32403 solver.cpp:406] Test net output #9: loss1/accuracy09 = 0.832
I0607 03:16:54.826256 32403 solver.cpp:406] Test net output #10: loss1/accuracy10 = 0.852
I0607 03:16:54.826268 32403 solver.cpp:406] Test net output #11: loss1/accuracy11 = 0.887
I0607 03:16:54.826280 32403 solver.cpp:406] Test net output #12: loss1/accuracy12 = 0.903
I0607 03:16:54.826293 32403 solver.cpp:406] Test net output #13: loss1/accuracy13 = 0.925
I0607 03:16:54.826304 32403 solver.cpp:406] Test net output #14: loss1/accuracy14 = 0.935
I0607 03:16:54.826316 32403 solver.cpp:406] Test net output #15: loss1/accuracy15 = 0.951
I0607 03:16:54.826328 32403 solver.cpp:406] Test net output #16: loss1/accuracy16 = 0.967
I0607 03:16:54.826340 32403 solver.cpp:406] Test net output #17: loss1/accuracy17 = 0.982
I0607 03:16:54.826352 32403 solver.cpp:406] Test net output #18: loss1/accuracy18 = 0.987
I0607 03:16:54.826364 32403 solver.cpp:406] Test net output #19: loss1/accuracy19 = 0.989
I0607 03:16:54.826376 32403 solver.cpp:406] Test net output #20: loss1/accuracy20 = 0.996
I0607 03:16:54.826388 32403 solver.cpp:406] Test net output #21: loss1/accuracy21 = 0.999
I0607 03:16:54.826400 32403 solver.cpp:406] Test net output #22: loss1/accuracy22 = 1
I0607 03:16:54.826412 32403 solver.cpp:406] Test net output #23: loss1/accuracy_incl_empty = 0.83782
I0607 03:16:54.826424 32403 solver.cpp:406] Test net output #24: loss1/accuracy_top3 = 0.787769
I0607 03:16:54.826441 32403 solver.cpp:406] Test net output #25: loss1/cross_entropy_loss = 1.60684 (* 0.3 = 0.482051 loss)
I0607 03:16:54.826455 32403 solver.cpp:406] Test net output #26: loss1/cross_entropy_loss_incl_empty = 0.599935 (* 0.3 = 0.179981 loss)
I0607 03:16:54.826470 32403 solver.cpp:406] Test net output #27: loss1/loss01 = 1.01852 (* 0.0272727 = 0.0277779 loss)
I0607 03:16:54.826484 32403 solver.cpp:406] Test net output #28: loss1/loss02 = 1.39121 (* 0.0272727 = 0.0379421 loss)
I0607 03:16:54.826498 32403 solver.cpp:406] Test net output #29: loss1/loss03 = 1.76863 (* 0.0272727 = 0.0482353 loss)
I0607 03:16:54.826511 32403 solver.cpp:406] Test net output #30: loss1/loss04 = 1.96236 (* 0.0272727 = 0.053519 loss)
I0607 03:16:54.826525 32403 solver.cpp:406] Test net output #31: loss1/loss05 = 1.76022 (* 0.0272727 = 0.0480061 loss)
I0607 03:16:54.826539 32403 solver.cpp:406] Test net output #32: loss1/loss06 = 1.28526 (* 0.0272727 = 0.0350526 loss)
I0607 03:16:54.826552 32403 solver.cpp:406] Test net output #33: loss1/loss07 = 0.857679 (* 0.0272727 = 0.0233912 loss)
I0607 03:16:54.826567 32403 solver.cpp:406] Test net output #34: loss1/loss08 = 0.708684 (* 0.0272727 = 0.0193277 loss)
I0607 03:16:54.826581 32403 solver.cpp:406] Test net output #35: loss1/loss09 = 0.606245 (* 0.0272727 = 0.0165339 loss)
I0607 03:16:54.826596 32403 solver.cpp:406] Test net output #36: loss1/loss10 = 0.529444 (* 0.0272727 = 0.0144394 loss)
I0607 03:16:54.826609 32403 solver.cpp:406] Test net output #37: loss1/loss11 = 0.459513 (* 0.0272727 = 0.0125322 loss)
I0607 03:16:54.826623 32403 solver.cpp:406] Test net output #38: loss1/loss12 = 0.381541 (* 0.0272727 = 0.0104057 loss)
I0607 03:16:54.826658 32403 solver.cpp:406] Test net output #39: loss1/loss13 = 0.281791 (* 0.0272727 = 0.0076852 loss)
I0607 03:16:54.826673 32403 solver.cpp:406] Test net output #40: loss1/loss14 = 0.24685 (* 0.0272727 = 0.00673228 loss)
I0607 03:16:54.826688 32403 solver.cpp:406] Test net output #41: loss1/loss15 = 0.188654 (* 0.0272727 = 0.0051451 loss)
I0607 03:16:54.826701 32403 solver.cpp:406] Test net output #42: loss1/loss16 = 0.149175 (* 0.0272727 = 0.00406841 loss)
I0607 03:16:54.826715 32403 solver.cpp:406] Test net output #43: loss1/loss17 = 0.093461 (* 0.0272727 = 0.00254894 loss)
I0607 03:16:54.826730 32403 solver.cpp:406] Test net output #44: loss1/loss18 = 0.0743486 (* 0.0272727 = 0.00202769 loss)
I0607 03:16:54.826743 32403 solver.cpp:406] Test net output #45: loss1/loss19 = 0.0631463 (* 0.0272727 = 0.00172217 loss)
I0607 03:16:54.826758 32403 solver.cpp:406] Test net output #46: loss1/loss20 = 0.0271806 (* 0.0272727 = 0.00074129 loss)
I0607 03:16:54.826772 32403 solver.cpp:406] Test net output #47: loss1/loss21 = 0.00901751 (* 0.0272727 = 0.000245932 loss)
I0607 03:16:54.826786 32403 solver.cpp:406] Test net output #48: loss1/loss22 = 0.00275754 (* 0.0272727 = 7.52058e-05 loss)
I0607 03:16:54.826798 32403 solver.cpp:406] Test net output #49: loss2/accuracy = 0.695676
I0607 03:16:54.826810 32403 solver.cpp:406] Test net output #50: loss2/accuracy01 = 0.866
I0607 03:16:54.826823 32403 solver.cpp:406] Test net output #51: loss2/accuracy02 = 0.821
I0607 03:16:54.826835 32403 solver.cpp:406] Test net output #52: loss2/accuracy03 = 0.726
I0607 03:16:54.826846 32403 solver.cpp:406] Test net output #53: loss2/accuracy04 = 0.605
I0607 03:16:54.826858 32403 solver.cpp:406] Test net output #54: loss2/accuracy05 = 0.605
I0607 03:16:54.826870 32403 solver.cpp:406] Test net output #55: loss2/accuracy06 = 0.734
I0607 03:16:54.826884 32403 solver.cpp:406] Test net output #56: loss2/accuracy07 = 0.789
I0607 03:16:54.826896 32403 solver.cpp:406] Test net output #57: loss2/accuracy08 = 0.823
I0607 03:16:54.826908 32403 solver.cpp:406] Test net output #58: loss2/accuracy09 = 0.853
I0607 03:16:54.826920 32403 solver.cpp:406] Test net output #59: loss2/accuracy10 = 0.865
I0607 03:16:54.826931 32403 solver.cpp:406] Test net output #60: loss2/accuracy11 = 0.893
I0607 03:16:54.826943 32403 solver.cpp:406] Test net output #61: loss2/accuracy12 = 0.906
I0607 03:16:54.826954 32403 solver.cpp:406] Test net output #62: loss2/accuracy13 = 0.928
I0607 03:16:54.826967 32403 solver.cpp:406] Test net output #63: loss2/accuracy14 = 0.936
I0607 03:16:54.826978 32403 solver.cpp:406] Test net output #64: loss2/accuracy15 = 0.952
I0607 03:16:54.826989 32403 solver.cpp:406] Test net output #65: loss2/accuracy16 = 0.968
I0607 03:16:54.827002 32403 solver.cpp:406] Test net output #66: loss2/accuracy17 = 0.982
I0607 03:16:54.827013 32403 solver.cpp:406] Test net output #67: loss2/accuracy18 = 0.987
I0607 03:16:54.827024 32403 solver.cpp:406] Test net output #68: loss2/accuracy19 = 0.989
I0607 03:16:54.827035 32403 solver.cpp:406] Test net output #69: loss2/accuracy20 = 0.996
I0607 03:16:54.827047 32403 solver.cpp:406] Test net output #70: loss2/accuracy21 = 0.999
I0607 03:16:54.827059 32403 solver.cpp:406] Test net output #71: loss2/accuracy22 = 1
I0607 03:16:54.827070 32403 solver.cpp:406] Test net output #72: loss2/accuracy_incl_empty = 0.876864
I0607 03:16:54.827081 32403 solver.cpp:406] Test net output #73: loss2/accuracy_top3 = 0.862608
I0607 03:16:54.827095 32403 solver.cpp:406] Test net output #74: loss2/cross_entropy_loss = 1.21677 (* 0.3 = 0.365031 loss)
I0607 03:16:54.827116 32403 solver.cpp:406] Test net output #75: loss2/cross_entropy_loss_incl_empty = 0.467345 (* 0.3 = 0.140204 loss)
I0607 03:16:54.827132 32403 solver.cpp:406] Test net output #76: loss2/loss01 = 0.723533 (* 0.0272727 = 0.0197327 loss)
I0607 03:16:54.827147 32403 solver.cpp:406] Test net output #77: loss2/loss02 = 0.923587 (* 0.0272727 = 0.0251887 loss)
I0607 03:16:54.827172 32403 solver.cpp:406] Test net output #78: loss2/loss03 = 1.12481 (* 0.0272727 = 0.0306767 loss)
I0607 03:16:54.827193 32403 solver.cpp:406] Test net output #79: loss2/loss04 = 1.36101 (* 0.0272727 = 0.0371184 loss)
I0607 03:16:54.827206 32403 solver.cpp:406] Test net output #80: loss2/loss05 = 1.32613 (* 0.0272727 = 0.036167 loss)
I0607 03:16:54.827220 32403 solver.cpp:406] Test net output #81: loss2/loss06 = 0.995467 (* 0.0272727 = 0.0271491 loss)
I0607 03:16:54.827234 32403 solver.cpp:406] Test net output #82: loss2/loss07 = 0.738083 (* 0.0272727 = 0.0201295 loss)
I0607 03:16:54.827247 32403 solver.cpp:406] Test net output #83: loss2/loss08 = 0.620743 (* 0.0272727 = 0.0169293 loss)
I0607 03:16:54.827263 32403 solver.cpp:406] Test net output #84: loss2/loss09 = 0.543799 (* 0.0272727 = 0.0148309 loss)
I0607 03:16:54.827277 32403 solver.cpp:406] Test net output #85: loss2/loss10 = 0.470618 (* 0.0272727 = 0.012835 loss)
I0607 03:16:54.827291 32403 solver.cpp:406] Test net output #86: loss2/loss11 = 0.408013 (* 0.0272727 = 0.0111276 loss)
I0607 03:16:54.827306 32403 solver.cpp:406] Test net output #87: loss2/loss12 = 0.334337 (* 0.0272727 = 0.00911827 loss)
I0607 03:16:54.827318 32403 solver.cpp:406] Test net output #88: loss2/loss13 = 0.247034 (* 0.0272727 = 0.00673728 loss)
I0607 03:16:54.827332 32403 solver.cpp:406] Test net output #89: loss2/loss14 = 0.220275 (* 0.0272727 = 0.00600749 loss)
I0607 03:16:54.827347 32403 solver.cpp:406] Test net output #90: loss2/loss15 = 0.166069 (* 0.0272727 = 0.00452916 loss)
I0607 03:16:54.827360 32403 solver.cpp:406] Test net output #91: loss2/loss16 = 0.135426 (* 0.0272727 = 0.00369343 loss)
I0607 03:16:54.827374 32403 solver.cpp:406] Test net output #92: loss2/loss17 = 0.0809676 (* 0.0272727 = 0.00220821 loss)
I0607 03:16:54.827389 32403 solver.cpp:406] Test net output #93: loss2/loss18 = 0.0616365 (* 0.0272727 = 0.001681 loss)
I0607 03:16:54.827402 32403 solver.cpp:406] Test net output #94: loss2/loss19 = 0.0616546 (* 0.0272727 = 0.00168149 loss)
I0607 03:16:54.827416 32403 solver.cpp:406] Test net output #95: loss2/loss20 = 0.0245711 (* 0.0272727 = 0.000670122 loss)
I0607 03:16:54.827430 32403 solver.cpp:406] Test net output #96: loss2/loss21 = 0.00785611 (* 0.0272727 = 0.000214258 loss)
I0607 03:16:54.827445 32403 solver.cpp:406] Test net output #97: loss2/loss22 = 0.00133014 (* 0.0272727 = 3.62765e-05 loss)
I0607 03:16:54.827457 32403 solver.cpp:406] Test net output #98: loss3/accuracy = 0.800934
I0607 03:16:54.827468 32403 solver.cpp:406] Test net output #99: loss3/accuracy01 = 0.877
I0607 03:16:54.827481 32403 solver.cpp:406] Test net output #100: loss3/accuracy02 = 0.856
I0607 03:16:54.827492 32403 solver.cpp:406] Test net output #101: loss3/accuracy03 = 0.837
I0607 03:16:54.827504 32403 solver.cpp:406] Test net output #102: loss3/accuracy04 = 0.813
I0607 03:16:54.827515 32403 solver.cpp:406] Test net output #103: loss3/accuracy05 = 0.805
I0607 03:16:54.827527 32403 solver.cpp:406] Test net output #104: loss3/accuracy06 = 0.832
I0607 03:16:54.827538 32403 solver.cpp:406] Test net output #105: loss3/accuracy07 = 0.871
I0607 03:16:54.827550 32403 solver.cpp:406] Test net output #106: loss3/accuracy08 = 0.886
I0607 03:16:54.827563 32403 solver.cpp:406] Test net output #107: loss3/accuracy09 = 0.89
I0607 03:16:54.827574 32403 solver.cpp:406] Test net output #108: loss3/accuracy10 = 0.896
I0607 03:16:54.827584 32403 solver.cpp:406] Test net output #109: loss3/accuracy11 = 0.916
I0607 03:16:54.827596 32403 solver.cpp:406] Test net output #110: loss3/accuracy12 = 0.929
I0607 03:16:54.827607 32403 solver.cpp:406] Test net output #111: loss3/accuracy13 = 0.942
I0607 03:16:54.827618 32403 solver.cpp:406] Test net output #112: loss3/accuracy14 = 0.952
I0607 03:16:54.827630 32403 solver.cpp:406] Test net output #113: loss3/accuracy15 = 0.962
I0607 03:16:54.827641 32403 solver.cpp:406] Test net output #114: loss3/accuracy16 = 0.971
I0607 03:16:54.827662 32403 solver.cpp:406] Test net output #115: loss3/accuracy17 = 0.986
I0607 03:16:54.827675 32403 solver.cpp:406] Test net output #116: loss3/accuracy18 = 0.989
I0607 03:16:54.827687 32403 solver.cpp:406] Test net output #117: loss3/accuracy19 = 0.99
I0607 03:16:54.827698 32403 solver.cpp:406] Test net output #118: loss3/accuracy20 = 0.996
I0607 03:16:54.827709 32403 solver.cpp:406] Test net output #119: loss3/accuracy21 = 0.999
I0607 03:16:54.827721 32403 solver.cpp:406] Test net output #120: loss3/accuracy22 = 1
I0607 03:16:54.827733 32403 solver.cpp:406] Test net output #121: loss3/accuracy_incl_empty = 0.916273
I0607 03:16:54.827744 32403 solver.cpp:406] Test net output #122: loss3/accuracy_top3 = 0.907201
I0607 03:16:54.827759 32403 solver.cpp:406] Test net output #123: loss3/cross_entropy_loss = 0.872759 (* 1 = 0.872759 loss)
I0607 03:16:54.827771 32403 solver.cpp:406] Test net output #124: loss3/cross_entropy_loss_incl_empty = 0.346821 (* 1 = 0.346821 loss)
I0607 03:16:54.827785 32403 solver.cpp:406] Test net output #125: loss3/loss01 = 0.626342 (* 0.0909091 = 0.0569401 loss)
I0607 03:16:54.827808 32403 solver.cpp:406] Test net output #126: loss3/loss02 = 0.769918 (* 0.0909091 = 0.0699925 loss)
I0607 03:16:54.827822 32403 solver.cpp:406] Test net output #127: loss3/loss03 = 0.814637 (* 0.0909091 = 0.0740579 loss)
I0607 03:16:54.827836 32403 solver.cpp:406] Test net output #128: loss3/loss04 = 0.804307 (* 0.0909091 = 0.0731188 loss)
I0607 03:16:54.827849 32403 solver.cpp:406] Test net output #129: loss3/loss05 = 0.798333 (* 0.0909091 = 0.0725758 loss)
I0607 03:16:54.827863 32403 solver.cpp:406] Test net output #130: loss3/loss06 = 0.671589 (* 0.0909091 = 0.0610536 loss)
I0607 03:16:54.827877 32403 solver.cpp:406] Test net output #131: loss3/loss07 = 0.503371 (* 0.0909091 = 0.045761 loss)
I0607 03:16:54.827890 32403 solver.cpp:406] Test net output #132: loss3/loss08 = 0.469274 (* 0.0909091 = 0.0426613 loss)
I0607 03:16:54.827904 32403 solver.cpp:406] Test net output #133: loss3/loss09 = 0.39353 (* 0.0909091 = 0.0357755 loss)
I0607 03:16:54.827919 32403 solver.cpp:406] Test net output #134: loss3/loss10 = 0.350874 (* 0.0909091 = 0.0318976 loss)
I0607 03:16:54.827934 32403 solver.cpp:406] Test net output #135: loss3/loss11 = 0.291257 (* 0.0909091 = 0.0264779 loss)
I0607 03:16:54.827949 32403 solver.cpp:406] Test net output #136: loss3/loss12 = 0.247464 (* 0.0909091 = 0.0224967 loss)
I0607 03:16:54.827962 32403 solver.cpp:406] Test net output #137: loss3/loss13 = 0.196724 (* 0.0909091 = 0.017884 loss)
I0607 03:16:54.827976 32403 solver.cpp:406] Test net output #138: loss3/loss14 = 0.158965 (* 0.0909091 = 0.0144514 loss)
I0607 03:16:54.827989 32403 solver.cpp:406] Test net output #139: loss3/loss15 = 0.13695 (* 0.0909091 = 0.01245 loss)
I0607 03:16:54.828007 32403 solver.cpp:406] Test net output #140: loss3/loss16 = 0.105032 (* 0.0909091 = 0.00954835 loss)
I0607 03:16:54.828022 32403 solver.cpp:406] Test net output #141: loss3/loss17 = 0.0629843 (* 0.0909091 = 0.00572584 loss)
I0607 03:16:54.828035 32403 solver.cpp:406] Test net output #142: loss3/loss18 = 0.0452659 (* 0.0909091 = 0.00411508 loss)
I0607 03:16:54.828049 32403 solver.cpp:406] Test net output #143: loss3/loss19 = 0.0400213 (* 0.0909091 = 0.0036383 loss)
I0607 03:16:54.828063 32403 solver.cpp:406] Test net output #144: loss3/loss20 = 0.0174593 (* 0.0909091 = 0.0015872 loss)
I0607 03:16:54.828078 32403 solver.cpp:406] Test net output #145: loss3/loss21 = 0.00616808 (* 0.0909091 = 0.000560734 loss)
I0607 03:16:54.828091 32403 solver.cpp:406] Test net output #146: loss3/loss22 = 0.000762891 (* 0.0909091 = 6.93537e-05 loss)
I0607 03:16:54.828104 32403 solver.cpp:406] Test net output #147: total_accuracy = 0.471
I0607 03:16:54.828114 32403 solver.cpp:406] Test net output #148: total_accuracy_not_rec = 0.473
I0607 03:16:54.828125 32403 solver.cpp:406] Test net output #149: total_confidence = 0.378651
I0607 03:16:54.828147 32403 solver.cpp:406] Test net output #150: total_confidence_not_rec = 0.361925
I0607 03:16:55.186205 32403 solver.cpp:229] Iteration 10000, loss = 3.97121
I0607 03:16:55.186277 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.534483
I0607 03:16:55.186296 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.625
I0607 03:16:55.186311 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.75
I0607 03:16:55.186323 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.75
I0607 03:16:55.186337 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.5
I0607 03:16:55.186349 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.625
I0607 03:16:55.186362 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.25
I0607 03:16:55.186375 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.875
I0607 03:16:55.186388 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 1
I0607 03:16:55.186401 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.875
I0607 03:16:55.186414 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.875
I0607 03:16:55.186426 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.875
I0607 03:16:55.186439 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 1
I0607 03:16:55.186451 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.875
I0607 03:16:55.186463 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.875
I0607 03:16:55.186476 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 0.875
I0607 03:16:55.186488 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 03:16:55.186501 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 03:16:55.186512 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 03:16:55.186524 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 03:16:55.186537 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 03:16:55.186548 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 03:16:55.186560 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 03:16:55.186573 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.829545
I0607 03:16:55.186584 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.827586
I0607 03:16:55.186601 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.3597 (* 0.3 = 0.407911 loss)
I0607 03:16:55.186615 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.541989 (* 0.3 = 0.162597 loss)
I0607 03:16:55.186630 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.773213 (* 0.0272727 = 0.0210876 loss)
I0607 03:16:55.186645 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 0.994352 (* 0.0272727 = 0.0271187 loss)
I0607 03:16:55.186658 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.28215 (* 0.0272727 = 0.0349677 loss)
I0607 03:16:55.186672 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.42828 (* 0.0272727 = 0.038953 loss)
I0607 03:16:55.186686 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.60731 (* 0.0272727 = 0.0438357 loss)
I0607 03:16:55.186700 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 2.00926 (* 0.0272727 = 0.054798 loss)
I0607 03:16:55.186717 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 0.714816 (* 0.0272727 = 0.019495 loss)
I0607 03:16:55.186733 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.279736 (* 0.0272727 = 0.00762916 loss)
I0607 03:16:55.186748 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.405548 (* 0.0272727 = 0.0110604 loss)
I0607 03:16:55.186761 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.358954 (* 0.0272727 = 0.00978965 loss)
I0607 03:16:55.186775 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.24632 (* 0.0272727 = 0.00671781 loss)
I0607 03:16:55.186825 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.16059 (* 0.0272727 = 0.00437971 loss)
I0607 03:16:55.186841 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.220528 (* 0.0272727 = 0.00601441 loss)
I0607 03:16:55.186856 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.261077 (* 0.0272727 = 0.00712028 loss)
I0607 03:16:55.186871 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.428592 (* 0.0272727 = 0.0116889 loss)
I0607 03:16:55.186884 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.0358096 (* 0.0272727 = 0.000976624 loss)
I0607 03:16:55.186899 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.0180504 (* 0.0272727 = 0.000492284 loss)
I0607 03:16:55.186913 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.0148732 (* 0.0272727 = 0.000405634 loss)
I0607 03:16:55.186928 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.0143094 (* 0.0272727 = 0.000390258 loss)
I0607 03:16:55.186941 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.00906476 (* 0.0272727 = 0.000247221 loss)
I0607 03:16:55.186956 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.0107737 (* 0.0272727 = 0.000293828 loss)
I0607 03:16:55.186970 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.0104209 (* 0.0272727 = 0.000284206 loss)
I0607 03:16:55.186982 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.637931
I0607 03:16:55.186995 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.625
I0607 03:16:55.187007 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.875
I0607 03:16:55.187019 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.75
I0607 03:16:55.187031 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.875
I0607 03:16:55.187043 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.625
I0607 03:16:55.187055 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.625
I0607 03:16:55.187067 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.875
I0607 03:16:55.187079 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 1
I0607 03:16:55.187091 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.875
I0607 03:16:55.187103 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 1
I0607 03:16:55.187114 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.875
I0607 03:16:55.187126 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.875
I0607 03:16:55.187139 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 1
I0607 03:16:55.187150 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.875
I0607 03:16:55.187163 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 0.875
I0607 03:16:55.187175 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 03:16:55.187186 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 03:16:55.187198 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 03:16:55.187211 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 03:16:55.187222 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 03:16:55.187233 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 03:16:55.187245 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 03:16:55.187257 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.857955
I0607 03:16:55.187269 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.931035
I0607 03:16:55.187283 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 0.954204 (* 0.3 = 0.286261 loss)
I0607 03:16:55.187296 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.399227 (* 0.3 = 0.119768 loss)
I0607 03:16:55.187321 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 1.23896 (* 0.0272727 = 0.0337899 loss)
I0607 03:16:55.187336 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.716329 (* 0.0272727 = 0.0195363 loss)
I0607 03:16:55.187350 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.677517 (* 0.0272727 = 0.0184777 loss)
I0607 03:16:55.187366 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 0.67551 (* 0.0272727 = 0.018423 loss)
I0607 03:16:55.187379 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.14201 (* 0.0272727 = 0.0311456 loss)
I0607 03:16:55.187393 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 0.889072 (* 0.0272727 = 0.0242474 loss)
I0607 03:16:55.187407 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.655459 (* 0.0272727 = 0.0178761 loss)
I0607 03:16:55.187422 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.215112 (* 0.0272727 = 0.00586669 loss)
I0607 03:16:55.187435 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.347734 (* 0.0272727 = 0.00948365 loss)
I0607 03:16:55.187449 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.229247 (* 0.0272727 = 0.00625219 loss)
I0607 03:16:55.187464 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.277905 (* 0.0272727 = 0.00757923 loss)
I0607 03:16:55.187479 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.275372 (* 0.0272727 = 0.00751016 loss)
I0607 03:16:55.187492 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.232148 (* 0.0272727 = 0.00633132 loss)
I0607 03:16:55.187506 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.268237 (* 0.0272727 = 0.00731555 loss)
I0607 03:16:55.187520 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.284615 (* 0.0272727 = 0.00776223 loss)
I0607 03:16:55.187536 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.113697 (* 0.0272727 = 0.00310082 loss)
I0607 03:16:55.187549 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.0153346 (* 0.0272727 = 0.000418216 loss)
I0607 03:16:55.187563 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.00801523 (* 0.0272727 = 0.000218597 loss)
I0607 03:16:55.187577 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.00279572 (* 0.0272727 = 7.6247e-05 loss)
I0607 03:16:55.187592 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.00297546 (* 0.0272727 = 8.1149e-05 loss)
I0607 03:16:55.187607 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.00431372 (* 0.0272727 = 0.000117647 loss)
I0607 03:16:55.187620 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 0.00431162 (* 0.0272727 = 0.00011759 loss)
I0607 03:16:55.187633 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.862069
I0607 03:16:55.187645 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 0.875
I0607 03:16:55.187657 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.875
I0607 03:16:55.187669 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.875
I0607 03:16:55.187680 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.875
I0607 03:16:55.187692 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.875
I0607 03:16:55.187705 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.875
I0607 03:16:55.187716 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.875
I0607 03:16:55.187727 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 1
I0607 03:16:55.187739 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 1
I0607 03:16:55.187752 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.875
I0607 03:16:55.187765 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 1
I0607 03:16:55.187778 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.875
I0607 03:16:55.187790 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 1
I0607 03:16:55.187813 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 0.875
I0607 03:16:55.187825 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 0.875
I0607 03:16:55.187840 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 03:16:55.187854 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 03:16:55.187865 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 03:16:55.187877 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 03:16:55.187888 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 03:16:55.187901 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 03:16:55.187912 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 03:16:55.187924 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.926136
I0607 03:16:55.187937 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.931035
I0607 03:16:55.187950 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.455314 (* 1 = 0.455314 loss)
I0607 03:16:55.187964 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.246633 (* 1 = 0.246633 loss)
I0607 03:16:55.187979 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.345752 (* 0.0909091 = 0.031432 loss)
I0607 03:16:55.187994 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.462392 (* 0.0909091 = 0.0420356 loss)
I0607 03:16:55.188007 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.290469 (* 0.0909091 = 0.0264063 loss)
I0607 03:16:55.188021 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.583244 (* 0.0909091 = 0.0530221 loss)
I0607 03:16:55.188035 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.330025 (* 0.0909091 = 0.0300023 loss)
I0607 03:16:55.188046 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.627279 (* 0.0909091 = 0.0570253 loss)
I0607 03:16:55.188062 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.374873 (* 0.0909091 = 0.0340794 loss)
I0607 03:16:55.188076 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.301378 (* 0.0909091 = 0.027398 loss)
I0607 03:16:55.188091 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.159527 (* 0.0909091 = 0.0145025 loss)
I0607 03:16:55.188105 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.28962 (* 0.0909091 = 0.0263291 loss)
I0607 03:16:55.188119 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.0584714 (* 0.0909091 = 0.00531558 loss)
I0607 03:16:55.188134 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.214285 (* 0.0909091 = 0.0194805 loss)
I0607 03:16:55.188148 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.111984 (* 0.0909091 = 0.0101804 loss)
I0607 03:16:55.188163 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.347034 (* 0.0909091 = 0.0315486 loss)
I0607 03:16:55.188176 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.241579 (* 0.0909091 = 0.0219617 loss)
I0607 03:16:55.188190 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.0508044 (* 0.0909091 = 0.00461858 loss)
I0607 03:16:55.188205 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.00257003 (* 0.0909091 = 0.000233639 loss)
I0607 03:16:55.188218 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.000797335 (* 0.0909091 = 7.2485e-05 loss)
I0607 03:16:55.188232 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.000310007 (* 0.0909091 = 2.81824e-05 loss)
I0607 03:16:55.188246 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.000214594 (* 0.0909091 = 1.95085e-05 loss)
I0607 03:16:55.188261 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.000251027 (* 0.0909091 = 2.28206e-05 loss)
I0607 03:16:55.188274 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 9.27216e-05 (* 0.0909091 = 8.42924e-06 loss)
I0607 03:16:55.188297 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.625
I0607 03:16:55.188311 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.75
I0607 03:16:55.188323 32403 solver.cpp:245] Train net output #149: total_confidence = 0.412852
I0607 03:16:55.188335 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.419265
I0607 03:16:55.188349 32403 sgd_solver.cpp:106] Iteration 10000, lr = 0.001
I0607 03:18:39.256669 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 38.6425 > 30) by scale factor 0.776347
I0607 03:19:23.344503 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.5415 > 30) by scale factor 0.982271
I0607 03:23:22.014837 32403 solver.cpp:229] Iteration 10500, loss = 3.90855
I0607 03:23:22.014991 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.482143
I0607 03:23:22.015013 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.875
I0607 03:23:22.015027 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.75
I0607 03:23:22.015040 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.5
I0607 03:23:22.015053 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.625
I0607 03:23:22.015067 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.5
I0607 03:23:22.015079 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.5
I0607 03:23:22.015092 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.75
I0607 03:23:22.015105 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.75
I0607 03:23:22.015118 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.75
I0607 03:23:22.015132 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.75
I0607 03:23:22.015146 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 1
I0607 03:23:22.015159 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 03:23:22.015172 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.875
I0607 03:23:22.015185 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.875
I0607 03:23:22.015197 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 03:23:22.015210 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 03:23:22.015223 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 03:23:22.015234 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 03:23:22.015246 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 03:23:22.015259 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 03:23:22.015270 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 03:23:22.015282 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 03:23:22.015295 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.835227
I0607 03:23:22.015306 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.714286
I0607 03:23:22.015329 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.7295 (* 0.3 = 0.518851 loss)
I0607 03:23:22.015344 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.594139 (* 0.3 = 0.178242 loss)
I0607 03:23:22.015359 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.609913 (* 0.0272727 = 0.016634 loss)
I0607 03:23:22.015374 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 0.89527 (* 0.0272727 = 0.0244164 loss)
I0607 03:23:22.015394 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.57645 (* 0.0272727 = 0.042994 loss)
I0607 03:23:22.015408 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.5254 (* 0.0272727 = 0.0416018 loss)
I0607 03:23:22.015424 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.71144 (* 0.0272727 = 0.0466757 loss)
I0607 03:23:22.015437 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.67709 (* 0.0272727 = 0.0457388 loss)
I0607 03:23:22.015451 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 0.586669 (* 0.0272727 = 0.0160001 loss)
I0607 03:23:22.015466 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.779831 (* 0.0272727 = 0.0212681 loss)
I0607 03:23:22.015480 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.957168 (* 0.0272727 = 0.0261046 loss)
I0607 03:23:22.015496 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 1.10215 (* 0.0272727 = 0.0300587 loss)
I0607 03:23:22.015509 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.281929 (* 0.0272727 = 0.00768898 loss)
I0607 03:23:22.015524 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.445836 (* 0.0272727 = 0.0121592 loss)
I0607 03:23:22.015563 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.344306 (* 0.0272727 = 0.00939016 loss)
I0607 03:23:22.015578 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.501696 (* 0.0272727 = 0.0136826 loss)
I0607 03:23:22.015594 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.104363 (* 0.0272727 = 0.00284627 loss)
I0607 03:23:22.015609 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.0445002 (* 0.0272727 = 0.00121364 loss)
I0607 03:23:22.015622 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.0255867 (* 0.0272727 = 0.00069782 loss)
I0607 03:23:22.015637 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.00548136 (* 0.0272727 = 0.000149492 loss)
I0607 03:23:22.015652 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.00239977 (* 0.0272727 = 6.54482e-05 loss)
I0607 03:23:22.015666 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.000537636 (* 0.0272727 = 1.46628e-05 loss)
I0607 03:23:22.015681 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.000466781 (* 0.0272727 = 1.27304e-05 loss)
I0607 03:23:22.015696 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.000840491 (* 0.0272727 = 2.29225e-05 loss)
I0607 03:23:22.015708 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.660714
I0607 03:23:22.015722 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.875
I0607 03:23:22.015733 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.875
I0607 03:23:22.015745 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.75
I0607 03:23:22.015758 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.875
I0607 03:23:22.015770 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.875
I0607 03:23:22.015782 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.625
I0607 03:23:22.015794 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.875
I0607 03:23:22.015806 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.875
I0607 03:23:22.015820 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.75
I0607 03:23:22.015831 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.75
I0607 03:23:22.015843 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 1
I0607 03:23:22.015856 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.875
I0607 03:23:22.015867 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.875
I0607 03:23:22.015882 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.875
I0607 03:23:22.015895 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 0.875
I0607 03:23:22.015908 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 0.875
I0607 03:23:22.015920 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 0.875
I0607 03:23:22.015933 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 03:23:22.015944 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 03:23:22.015956 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 03:23:22.015969 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 03:23:22.015981 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 03:23:22.015992 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.869318
I0607 03:23:22.016012 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.785714
I0607 03:23:22.016026 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.28669 (* 0.3 = 0.386008 loss)
I0607 03:23:22.016042 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.466415 (* 0.3 = 0.139925 loss)
I0607 03:23:22.016057 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.214796 (* 0.0272727 = 0.00585807 loss)
I0607 03:23:22.016078 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.259477 (* 0.0272727 = 0.00707664 loss)
I0607 03:23:22.016104 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.863918 (* 0.0272727 = 0.0235614 loss)
I0607 03:23:22.016120 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 0.825104 (* 0.0272727 = 0.0225028 loss)
I0607 03:23:22.016134 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 0.714844 (* 0.0272727 = 0.0194958 loss)
I0607 03:23:22.016149 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.2149 (* 0.0272727 = 0.0331336 loss)
I0607 03:23:22.016163 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.745451 (* 0.0272727 = 0.0203305 loss)
I0607 03:23:22.016178 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.559204 (* 0.0272727 = 0.015251 loss)
I0607 03:23:22.016192 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 1.2137 (* 0.0272727 = 0.033101 loss)
I0607 03:23:22.016206 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.92629 (* 0.0272727 = 0.0252624 loss)
I0607 03:23:22.016221 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.128694 (* 0.0272727 = 0.00350983 loss)
I0607 03:23:22.016235 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.401473 (* 0.0272727 = 0.0109493 loss)
I0607 03:23:22.016249 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.39355 (* 0.0272727 = 0.0107332 loss)
I0607 03:23:22.016263 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.533921 (* 0.0272727 = 0.0145615 loss)
I0607 03:23:22.016278 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.283827 (* 0.0272727 = 0.00774073 loss)
I0607 03:23:22.016293 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.198396 (* 0.0272727 = 0.00541079 loss)
I0607 03:23:22.016307 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.196139 (* 0.0272727 = 0.00534923 loss)
I0607 03:23:22.016321 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.0519494 (* 0.0272727 = 0.0014168 loss)
I0607 03:23:22.016336 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.00204938 (* 0.0272727 = 5.58923e-05 loss)
I0607 03:23:22.016351 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.00036164 (* 0.0272727 = 9.8629e-06 loss)
I0607 03:23:22.016366 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 4.10489e-05 (* 0.0272727 = 1.11951e-06 loss)
I0607 03:23:22.016381 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 1.2577e-05 (* 0.0272727 = 3.43009e-07 loss)
I0607 03:23:22.016392 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.767857
I0607 03:23:22.016405 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 03:23:22.016417 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 1
I0607 03:23:22.016429 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 1
I0607 03:23:22.016441 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 1
I0607 03:23:22.016454 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.875
I0607 03:23:22.016466 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.75
I0607 03:23:22.016479 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.875
I0607 03:23:22.016491 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.875
I0607 03:23:22.016504 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.75
I0607 03:23:22.016515 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.75
I0607 03:23:22.016526 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.875
I0607 03:23:22.016538 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.875
I0607 03:23:22.016551 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 0.875
I0607 03:23:22.016562 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 0.875
I0607 03:23:22.016574 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 0.875
I0607 03:23:22.016597 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 03:23:22.016610 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 03:23:22.016623 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 03:23:22.016635 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 03:23:22.016647 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 03:23:22.016659 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 03:23:22.016671 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 03:23:22.016683 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.914773
I0607 03:23:22.016695 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.892857
I0607 03:23:22.016710 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.705158 (* 1 = 0.705158 loss)
I0607 03:23:22.016723 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.262172 (* 1 = 0.262172 loss)
I0607 03:23:22.016738 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.0551762 (* 0.0909091 = 0.00501602 loss)
I0607 03:23:22.016753 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.0128606 (* 0.0909091 = 0.00116915 loss)
I0607 03:23:22.016769 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.0367904 (* 0.0909091 = 0.00334458 loss)
I0607 03:23:22.016784 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.0904844 (* 0.0909091 = 0.00822585 loss)
I0607 03:23:22.016798 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.351255 (* 0.0909091 = 0.0319323 loss)
I0607 03:23:22.016813 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.838104 (* 0.0909091 = 0.0761913 loss)
I0607 03:23:22.016827 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.46799 (* 0.0909091 = 0.0425445 loss)
I0607 03:23:22.016842 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.313522 (* 0.0909091 = 0.028502 loss)
I0607 03:23:22.016856 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.627104 (* 0.0909091 = 0.0570095 loss)
I0607 03:23:22.016870 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.785746 (* 0.0909091 = 0.0714315 loss)
I0607 03:23:22.016885 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.358961 (* 0.0909091 = 0.0326328 loss)
I0607 03:23:22.016903 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.240943 (* 0.0909091 = 0.0219039 loss)
I0607 03:23:22.016917 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.286208 (* 0.0909091 = 0.0260189 loss)
I0607 03:23:22.016934 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.297937 (* 0.0909091 = 0.0270852 loss)
I0607 03:23:22.016958 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.18259 (* 0.0909091 = 0.0165991 loss)
I0607 03:23:22.016973 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.0187835 (* 0.0909091 = 0.00170759 loss)
I0607 03:23:22.016988 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.0112527 (* 0.0909091 = 0.00102297 loss)
I0607 03:23:22.017001 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.00399184 (* 0.0909091 = 0.000362894 loss)
I0607 03:23:22.017015 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.00270876 (* 0.0909091 = 0.000246251 loss)
I0607 03:23:22.017033 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.000318984 (* 0.0909091 = 2.89985e-05 loss)
I0607 03:23:22.017051 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.000109461 (* 0.0909091 = 9.95103e-06 loss)
I0607 03:23:22.017066 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 2.81797e-05 (* 0.0909091 = 2.56179e-06 loss)
I0607 03:23:22.017078 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.625
I0607 03:23:22.017093 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.75
I0607 03:23:22.017127 32403 solver.cpp:245] Train net output #149: total_confidence = 0.59676
I0607 03:23:22.017143 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.636784
I0607 03:23:22.017158 32403 sgd_solver.cpp:106] Iteration 10500, lr = 0.001
I0607 03:23:40.958302 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 41.666 > 30) by scale factor 0.720012
I0607 03:25:41.647860 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.9309 > 30) by scale factor 0.93953
I0607 03:28:57.483088 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.8802 > 30) by scale factor 0.885472
I0607 03:29:06.760200 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.2673 > 30) by scale factor 0.901786
I0607 03:29:27.635783 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.4199 > 30) by scale factor 0.954808
I0607 03:29:48.918023 32403 solver.cpp:229] Iteration 11000, loss = 4.01932
I0607 03:29:48.918115 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.417722
I0607 03:29:48.918135 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.75
I0607 03:29:48.918149 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.625
I0607 03:29:48.918164 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.25
I0607 03:29:48.918176 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.25
I0607 03:29:48.918190 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.25
I0607 03:29:48.918203 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.5
I0607 03:29:48.918216 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.75
I0607 03:29:48.918229 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.625
I0607 03:29:48.918242 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.625
I0607 03:29:48.918256 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.625
I0607 03:29:48.918268 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.625
I0607 03:29:48.918280 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.75
I0607 03:29:48.918293 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.625
I0607 03:29:48.918306 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.625
I0607 03:29:48.918318 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 0.625
I0607 03:29:48.918330 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 0.75
I0607 03:29:48.918342 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 0.875
I0607 03:29:48.918355 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 03:29:48.918367 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 03:29:48.918380 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 03:29:48.918392 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 03:29:48.918404 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 03:29:48.918416 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.732955
I0607 03:29:48.918428 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.632911
I0607 03:29:48.918445 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 2.01807 (* 0.3 = 0.605422 loss)
I0607 03:29:48.918460 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.950835 (* 0.3 = 0.28525 loss)
I0607 03:29:48.918475 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.955931 (* 0.0272727 = 0.0260708 loss)
I0607 03:29:48.918489 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.23081 (* 0.0272727 = 0.0335676 loss)
I0607 03:29:48.918503 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 2.08967 (* 0.0272727 = 0.0569909 loss)
I0607 03:29:48.918517 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.95612 (* 0.0272727 = 0.0533487 loss)
I0607 03:29:48.918532 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 2.94197 (* 0.0272727 = 0.0802356 loss)
I0607 03:29:48.918546 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.80546 (* 0.0272727 = 0.0492397 loss)
I0607 03:29:48.918560 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 1.16352 (* 0.0272727 = 0.0317322 loss)
I0607 03:29:48.918575 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 1.28577 (* 0.0272727 = 0.0350664 loss)
I0607 03:29:48.918588 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 1.4871 (* 0.0272727 = 0.0405573 loss)
I0607 03:29:48.918602 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 1.5412 (* 0.0272727 = 0.0420327 loss)
I0607 03:29:48.918617 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 1.23048 (* 0.0272727 = 0.0335586 loss)
I0607 03:29:48.918630 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 1.17406 (* 0.0272727 = 0.0320199 loss)
I0607 03:29:48.918684 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 1.06194 (* 0.0272727 = 0.0289621 loss)
I0607 03:29:48.918699 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.982836 (* 0.0272727 = 0.0268046 loss)
I0607 03:29:48.918712 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 1.03237 (* 0.0272727 = 0.0281555 loss)
I0607 03:29:48.918727 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 1.0656 (* 0.0272727 = 0.0290619 loss)
I0607 03:29:48.918741 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.712298 (* 0.0272727 = 0.0194263 loss)
I0607 03:29:48.918756 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.0331692 (* 0.0272727 = 0.000904614 loss)
I0607 03:29:48.918773 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.0312369 (* 0.0272727 = 0.000851915 loss)
I0607 03:29:48.918788 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.0225908 (* 0.0272727 = 0.000616114 loss)
I0607 03:29:48.918803 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.0160991 (* 0.0272727 = 0.000439068 loss)
I0607 03:29:48.918818 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.0126507 (* 0.0272727 = 0.000345018 loss)
I0607 03:29:48.918830 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.417722
I0607 03:29:48.918846 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.875
I0607 03:29:48.918860 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.75
I0607 03:29:48.918872 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.625
I0607 03:29:48.918884 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.625
I0607 03:29:48.918896 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.375
I0607 03:29:48.918910 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.375
I0607 03:29:48.918921 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.5
I0607 03:29:48.918933 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.625
I0607 03:29:48.918946 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.625
I0607 03:29:48.918958 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.5
I0607 03:29:48.918970 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.625
I0607 03:29:48.918982 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.75
I0607 03:29:48.918994 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.625
I0607 03:29:48.919006 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.75
I0607 03:29:48.919018 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 0.75
I0607 03:29:48.919031 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 0.75
I0607 03:29:48.919044 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 0.875
I0607 03:29:48.919055 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 03:29:48.919067 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 03:29:48.919080 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 03:29:48.919091 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 03:29:48.919103 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 03:29:48.919116 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.738636
I0607 03:29:48.919127 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.708861
I0607 03:29:48.919142 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.83021 (* 0.3 = 0.549062 loss)
I0607 03:29:48.919157 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.840891 (* 0.3 = 0.252267 loss)
I0607 03:29:48.919170 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.405787 (* 0.0272727 = 0.0110669 loss)
I0607 03:29:48.919196 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 1.22365 (* 0.0272727 = 0.0333722 loss)
I0607 03:29:48.919212 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 1.54302 (* 0.0272727 = 0.0420825 loss)
I0607 03:29:48.919226 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.3355 (* 0.0272727 = 0.0364227 loss)
I0607 03:29:48.919240 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 2.0218 (* 0.0272727 = 0.0551399 loss)
I0607 03:29:48.919255 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 2.22227 (* 0.0272727 = 0.0606073 loss)
I0607 03:29:48.919270 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 1.45988 (* 0.0272727 = 0.039815 loss)
I0607 03:29:48.919283 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 1.4129 (* 0.0272727 = 0.0385336 loss)
I0607 03:29:48.919297 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 1.43461 (* 0.0272727 = 0.0391258 loss)
I0607 03:29:48.919312 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 1.33216 (* 0.0272727 = 0.0363316 loss)
I0607 03:29:48.919324 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 1.04329 (* 0.0272727 = 0.0284535 loss)
I0607 03:29:48.919338 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 1.13987 (* 0.0272727 = 0.0310874 loss)
I0607 03:29:48.919353 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 1.0383 (* 0.0272727 = 0.0283172 loss)
I0607 03:29:48.919366 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.796259 (* 0.0272727 = 0.0217162 loss)
I0607 03:29:48.919380 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.978797 (* 0.0272727 = 0.0266945 loss)
I0607 03:29:48.919394 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.863021 (* 0.0272727 = 0.0235369 loss)
I0607 03:29:48.919409 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.760071 (* 0.0272727 = 0.0207292 loss)
I0607 03:29:48.919423 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.00687756 (* 0.0272727 = 0.00018757 loss)
I0607 03:29:48.919438 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.0040512 (* 0.0272727 = 0.000110487 loss)
I0607 03:29:48.919452 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.00147192 (* 0.0272727 = 4.01431e-05 loss)
I0607 03:29:48.919467 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.00106255 (* 0.0272727 = 2.89788e-05 loss)
I0607 03:29:48.919482 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 0.000566685 (* 0.0272727 = 1.54551e-05 loss)
I0607 03:29:48.919495 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.531646
I0607 03:29:48.919507 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 0.875
I0607 03:29:48.919519 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.875
I0607 03:29:48.919531 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.75
I0607 03:29:48.919543 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.875
I0607 03:29:48.919555 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.5
I0607 03:29:48.919569 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.5
I0607 03:29:48.919580 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.875
I0607 03:29:48.919592 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.75
I0607 03:29:48.919605 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.625
I0607 03:29:48.919616 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.625
I0607 03:29:48.919628 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.75
I0607 03:29:48.919641 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.75
I0607 03:29:48.919652 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 0.75
I0607 03:29:48.919664 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 0.75
I0607 03:29:48.919677 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 0.625
I0607 03:29:48.919698 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 0.75
I0607 03:29:48.919713 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 0.875
I0607 03:29:48.919724 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 03:29:48.919734 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 03:29:48.919741 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 03:29:48.919754 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 03:29:48.919766 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 03:29:48.919777 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.784091
I0607 03:29:48.919790 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.797468
I0607 03:29:48.919805 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 1.37732 (* 1 = 1.37732 loss)
I0607 03:29:48.919818 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.645853 (* 1 = 0.645853 loss)
I0607 03:29:48.919836 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.445041 (* 0.0909091 = 0.0404582 loss)
I0607 03:29:48.919857 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.417466 (* 0.0909091 = 0.0379515 loss)
I0607 03:29:48.919872 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.635279 (* 0.0909091 = 0.0577527 loss)
I0607 03:29:48.919885 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.484357 (* 0.0909091 = 0.0440324 loss)
I0607 03:29:48.919910 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 1.76037 (* 0.0909091 = 0.160033 loss)
I0607 03:29:48.919925 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 1.92239 (* 0.0909091 = 0.174762 loss)
I0607 03:29:48.919939 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.715587 (* 0.0909091 = 0.0650534 loss)
I0607 03:29:48.919953 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 1.00976 (* 0.0909091 = 0.0917967 loss)
I0607 03:29:48.919967 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 1.20875 (* 0.0909091 = 0.109886 loss)
I0607 03:29:48.919981 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 1.24187 (* 0.0909091 = 0.112897 loss)
I0607 03:29:48.919996 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.669586 (* 0.0909091 = 0.0608714 loss)
I0607 03:29:48.920009 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.741526 (* 0.0909091 = 0.0674115 loss)
I0607 03:29:48.920023 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.916135 (* 0.0909091 = 0.083285 loss)
I0607 03:29:48.920038 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.758382 (* 0.0909091 = 0.0689438 loss)
I0607 03:29:48.920052 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.95281 (* 0.0909091 = 0.0866191 loss)
I0607 03:29:48.920066 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.741632 (* 0.0909091 = 0.0674211 loss)
I0607 03:29:48.920080 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.831933 (* 0.0909091 = 0.0756302 loss)
I0607 03:29:48.920094 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.00673043 (* 0.0909091 = 0.000611857 loss)
I0607 03:29:48.920109 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.00190384 (* 0.0909091 = 0.000173077 loss)
I0607 03:29:48.920123 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.00051504 (* 0.0909091 = 4.68218e-05 loss)
I0607 03:29:48.920138 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.000318578 (* 0.0909091 = 2.89617e-05 loss)
I0607 03:29:48.920152 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 0.000160849 (* 0.0909091 = 1.46227e-05 loss)
I0607 03:29:48.920164 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.25
I0607 03:29:48.920176 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.25
I0607 03:29:48.920198 32403 solver.cpp:245] Train net output #149: total_confidence = 0.256377
I0607 03:29:48.920212 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.306398
I0607 03:29:48.920225 32403 sgd_solver.cpp:106] Iteration 11000, lr = 0.001
I0607 03:29:49.280133 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 44.2454 > 30) by scale factor 0.678036
I0607 03:31:07.274080 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.101 > 30) by scale factor 0.879741
I0607 03:32:15.329478 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.809 > 30) by scale factor 0.914382
I0607 03:34:05.126184 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.7721 > 30) by scale factor 0.944223
I0607 03:35:07.034147 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 35.0061 > 30) by scale factor 0.856993
I0607 03:35:19.396173 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.6461 > 30) by scale factor 0.978918
I0607 03:36:13.541129 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.5528 > 30) by scale factor 0.950787
I0607 03:36:16.267488 32403 solver.cpp:229] Iteration 11500, loss = 4.04779
I0607 03:36:16.267550 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.530612
I0607 03:36:16.267568 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.625
I0607 03:36:16.267583 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.5
I0607 03:36:16.267596 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.375
I0607 03:36:16.267609 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.375
I0607 03:36:16.267622 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.25
I0607 03:36:16.267635 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 1
I0607 03:36:16.267648 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.5
I0607 03:36:16.267662 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.875
I0607 03:36:16.267674 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 1
I0607 03:36:16.267686 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 1
I0607 03:36:16.267702 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 1
I0607 03:36:16.267715 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 1
I0607 03:36:16.267729 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 1
I0607 03:36:16.267740 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 1
I0607 03:36:16.267752 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 03:36:16.267763 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 03:36:16.267776 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 03:36:16.267787 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 03:36:16.267799 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 03:36:16.267812 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 03:36:16.267823 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 03:36:16.267835 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 03:36:16.267848 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.863636
I0607 03:36:16.267859 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.77551
I0607 03:36:16.267876 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.50393 (* 0.3 = 0.451179 loss)
I0607 03:36:16.267891 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.477748 (* 0.3 = 0.143324 loss)
I0607 03:36:16.267906 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 1.0655 (* 0.0272727 = 0.029059 loss)
I0607 03:36:16.267920 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.32717 (* 0.0272727 = 0.0361956 loss)
I0607 03:36:16.267935 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 2.02289 (* 0.0272727 = 0.0551696 loss)
I0607 03:36:16.267949 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.90754 (* 0.0272727 = 0.0520237 loss)
I0607 03:36:16.267963 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.91236 (* 0.0272727 = 0.0521553 loss)
I0607 03:36:16.267977 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 0.474391 (* 0.0272727 = 0.0129379 loss)
I0607 03:36:16.267992 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 1.15215 (* 0.0272727 = 0.0314221 loss)
I0607 03:36:16.268005 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.532812 (* 0.0272727 = 0.0145312 loss)
I0607 03:36:16.268020 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.0762567 (* 0.0272727 = 0.00207973 loss)
I0607 03:36:16.268034 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.0478448 (* 0.0272727 = 0.00130486 loss)
I0607 03:36:16.268049 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.0405901 (* 0.0272727 = 0.001107 loss)
I0607 03:36:16.268064 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.0283281 (* 0.0272727 = 0.000772583 loss)
I0607 03:36:16.268108 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.032061 (* 0.0272727 = 0.000874391 loss)
I0607 03:36:16.268124 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.0319701 (* 0.0272727 = 0.000871912 loss)
I0607 03:36:16.268139 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.0215701 (* 0.0272727 = 0.000588277 loss)
I0607 03:36:16.268153 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.0177602 (* 0.0272727 = 0.000484368 loss)
I0607 03:36:16.268168 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.0162973 (* 0.0272727 = 0.00044447 loss)
I0607 03:36:16.268182 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.0149148 (* 0.0272727 = 0.000406768 loss)
I0607 03:36:16.268196 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.0116814 (* 0.0272727 = 0.000318583 loss)
I0607 03:36:16.268211 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.0121899 (* 0.0272727 = 0.000332452 loss)
I0607 03:36:16.268225 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.0136485 (* 0.0272727 = 0.000372231 loss)
I0607 03:36:16.268244 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.0132207 (* 0.0272727 = 0.000360564 loss)
I0607 03:36:16.268257 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.714286
I0607 03:36:16.268270 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.875
I0607 03:36:16.268282 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.75
I0607 03:36:16.268296 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.75
I0607 03:36:16.268307 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.25
I0607 03:36:16.268319 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.75
I0607 03:36:16.268332 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.75
I0607 03:36:16.268344 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.875
I0607 03:36:16.268357 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.875
I0607 03:36:16.268368 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 1
I0607 03:36:16.268380 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 1
I0607 03:36:16.268393 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 1
I0607 03:36:16.268404 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 1
I0607 03:36:16.268415 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 1
I0607 03:36:16.268427 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 1
I0607 03:36:16.268438 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 03:36:16.268450 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 03:36:16.268462 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 03:36:16.268474 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 03:36:16.268486 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 03:36:16.268498 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 03:36:16.268509 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 03:36:16.268522 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 03:36:16.268533 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.909091
I0607 03:36:16.268546 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.857143
I0607 03:36:16.268560 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.07907 (* 0.3 = 0.323722 loss)
I0607 03:36:16.268574 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.380507 (* 0.3 = 0.114152 loss)
I0607 03:36:16.268589 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.647593 (* 0.0272727 = 0.0176616 loss)
I0607 03:36:16.268604 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 1.21158 (* 0.0272727 = 0.033043 loss)
I0607 03:36:16.268628 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 1.12524 (* 0.0272727 = 0.0306884 loss)
I0607 03:36:16.268645 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.77275 (* 0.0272727 = 0.0483479 loss)
I0607 03:36:16.268658 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.36477 (* 0.0272727 = 0.0372211 loss)
I0607 03:36:16.268672 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 0.805796 (* 0.0272727 = 0.0219763 loss)
I0607 03:36:16.268687 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.56028 (* 0.0272727 = 0.0152804 loss)
I0607 03:36:16.268700 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.445925 (* 0.0272727 = 0.0121616 loss)
I0607 03:36:16.268715 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.117776 (* 0.0272727 = 0.00321206 loss)
I0607 03:36:16.268729 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.0952936 (* 0.0272727 = 0.00259892 loss)
I0607 03:36:16.268743 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.0681417 (* 0.0272727 = 0.00185841 loss)
I0607 03:36:16.268761 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.0545473 (* 0.0272727 = 0.00148765 loss)
I0607 03:36:16.268775 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.0635225 (* 0.0272727 = 0.00173243 loss)
I0607 03:36:16.268790 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.0507559 (* 0.0272727 = 0.00138425 loss)
I0607 03:36:16.268805 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.0512286 (* 0.0272727 = 0.00139714 loss)
I0607 03:36:16.268818 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.0728092 (* 0.0272727 = 0.00198571 loss)
I0607 03:36:16.268832 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.0612726 (* 0.0272727 = 0.00167107 loss)
I0607 03:36:16.268846 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.0565479 (* 0.0272727 = 0.00154222 loss)
I0607 03:36:16.268860 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.0630371 (* 0.0272727 = 0.00171919 loss)
I0607 03:36:16.268875 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.0699333 (* 0.0272727 = 0.00190727 loss)
I0607 03:36:16.268889 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.0716758 (* 0.0272727 = 0.00195479 loss)
I0607 03:36:16.268903 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 0.0789325 (* 0.0272727 = 0.0021527 loss)
I0607 03:36:16.268913 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.877551
I0607 03:36:16.268920 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 0.875
I0607 03:36:16.268934 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.875
I0607 03:36:16.268946 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.875
I0607 03:36:16.268959 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.875
I0607 03:36:16.268970 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.875
I0607 03:36:16.268981 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.875
I0607 03:36:16.268993 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 1
I0607 03:36:16.269006 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.875
I0607 03:36:16.269017 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 1
I0607 03:36:16.269029 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 1
I0607 03:36:16.269040 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 1
I0607 03:36:16.269052 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 1
I0607 03:36:16.269063 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 1
I0607 03:36:16.269075 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 03:36:16.269086 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 03:36:16.269109 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 03:36:16.269140 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 03:36:16.269155 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 03:36:16.269166 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 03:36:16.269178 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 03:36:16.269191 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 03:36:16.269202 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 03:36:16.269213 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.965909
I0607 03:36:16.269225 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.897959
I0607 03:36:16.269239 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.552946 (* 1 = 0.552946 loss)
I0607 03:36:16.269253 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.190102 (* 1 = 0.190102 loss)
I0607 03:36:16.269268 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.438649 (* 0.0909091 = 0.0398772 loss)
I0607 03:36:16.269282 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.677686 (* 0.0909091 = 0.0616078 loss)
I0607 03:36:16.269299 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.569413 (* 0.0909091 = 0.0517649 loss)
I0607 03:36:16.269315 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.56088 (* 0.0909091 = 0.0509891 loss)
I0607 03:36:16.269328 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.712884 (* 0.0909091 = 0.0648076 loss)
I0607 03:36:16.269342 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.296726 (* 0.0909091 = 0.0269751 loss)
I0607 03:36:16.269356 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.192093 (* 0.0909091 = 0.017463 loss)
I0607 03:36:16.269371 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.317314 (* 0.0909091 = 0.0288467 loss)
I0607 03:36:16.269385 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.0451774 (* 0.0909091 = 0.00410704 loss)
I0607 03:36:16.269399 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.0323243 (* 0.0909091 = 0.00293857 loss)
I0607 03:36:16.269413 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.0234862 (* 0.0909091 = 0.00213511 loss)
I0607 03:36:16.269426 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.0132355 (* 0.0909091 = 0.00120323 loss)
I0607 03:36:16.269440 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.00973947 (* 0.0909091 = 0.000885407 loss)
I0607 03:36:16.269455 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.00670726 (* 0.0909091 = 0.000609751 loss)
I0607 03:36:16.269469 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.0066543 (* 0.0909091 = 0.000604936 loss)
I0607 03:36:16.269484 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.00537681 (* 0.0909091 = 0.000488801 loss)
I0607 03:36:16.269497 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.00801932 (* 0.0909091 = 0.000729029 loss)
I0607 03:36:16.269512 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.00821303 (* 0.0909091 = 0.000746639 loss)
I0607 03:36:16.269526 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.00716921 (* 0.0909091 = 0.000651746 loss)
I0607 03:36:16.269541 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.00611903 (* 0.0909091 = 0.000556276 loss)
I0607 03:36:16.269554 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.00583658 (* 0.0909091 = 0.000530598 loss)
I0607 03:36:16.269568 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 0.00757796 (* 0.0909091 = 0.000688906 loss)
I0607 03:36:16.269580 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.75
I0607 03:36:16.269593 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.625
I0607 03:36:16.269615 32403 solver.cpp:245] Train net output #149: total_confidence = 0.468218
I0607 03:36:16.269629 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.425656
I0607 03:36:16.269639 32403 sgd_solver.cpp:106] Iteration 11500, lr = 0.001
I0607 03:36:58.432296 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.0835 > 30) by scale factor 0.93506
I0607 03:37:59.517907 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 48.7806 > 30) by scale factor 0.614999
I0607 03:38:51.371721 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 35.8116 > 30) by scale factor 0.837719
I0607 03:39:48.639467 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.5671 > 30) by scale factor 0.981448
I0607 03:41:20.693632 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.4553 > 30) by scale factor 0.985049
I0607 03:42:16.360224 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.0461 > 30) by scale factor 0.809801
I0607 03:42:18.678802 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.4894 > 30) by scale factor 0.869832
I0607 03:42:43.062341 32403 solver.cpp:229] Iteration 12000, loss = 4.03031
I0607 03:42:43.062418 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.644444
I0607 03:42:43.062438 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.75
I0607 03:42:43.062453 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.625
I0607 03:42:43.062468 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.75
I0607 03:42:43.062480 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.375
I0607 03:42:43.062494 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.5
I0607 03:42:43.062506 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.875
I0607 03:42:43.062520 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.875
I0607 03:42:43.062532 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.875
I0607 03:42:43.062546 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 1
I0607 03:42:43.062557 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.875
I0607 03:42:43.062572 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 1
I0607 03:42:43.062583 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 03:42:43.062597 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.875
I0607 03:42:43.062626 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 1
I0607 03:42:43.062647 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 03:42:43.062661 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 03:42:43.062674 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 03:42:43.062685 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 03:42:43.062697 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 03:42:43.062710 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 03:42:43.062722 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 03:42:43.062734 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 03:42:43.062747 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.903409
I0607 03:42:43.062762 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.844444
I0607 03:42:43.062778 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.09334 (* 0.3 = 0.328001 loss)
I0607 03:42:43.062793 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.32581 (* 0.3 = 0.097743 loss)
I0607 03:42:43.062808 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.564003 (* 0.0272727 = 0.0153819 loss)
I0607 03:42:43.062824 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.12203 (* 0.0272727 = 0.0306009 loss)
I0607 03:42:43.062837 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.28982 (* 0.0272727 = 0.035177 loss)
I0607 03:42:43.062851 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.91903 (* 0.0272727 = 0.0523373 loss)
I0607 03:42:43.062866 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.3536 (* 0.0272727 = 0.0369162 loss)
I0607 03:42:43.062880 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 0.414396 (* 0.0272727 = 0.0113017 loss)
I0607 03:42:43.062896 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 0.335107 (* 0.0272727 = 0.00913927 loss)
I0607 03:42:43.062909 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.245525 (* 0.0272727 = 0.00669613 loss)
I0607 03:42:43.062924 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.157534 (* 0.0272727 = 0.00429638 loss)
I0607 03:42:43.062939 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.363742 (* 0.0272727 = 0.00992023 loss)
I0607 03:42:43.062953 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.206541 (* 0.0272727 = 0.00563293 loss)
I0607 03:42:43.063007 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.355888 (* 0.0272727 = 0.00970604 loss)
I0607 03:42:43.063024 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.362082 (* 0.0272727 = 0.00987496 loss)
I0607 03:42:43.063038 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.016316 (* 0.0272727 = 0.000444983 loss)
I0607 03:42:43.063053 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.00539715 (* 0.0272727 = 0.000147195 loss)
I0607 03:42:43.063067 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.00359595 (* 0.0272727 = 9.80714e-05 loss)
I0607 03:42:43.063082 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.000338654 (* 0.0272727 = 9.23602e-06 loss)
I0607 03:42:43.063097 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 3.4195e-05 (* 0.0272727 = 9.32592e-07 loss)
I0607 03:42:43.063112 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 9.99898e-06 (* 0.0272727 = 2.72699e-07 loss)
I0607 03:42:43.063127 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 1.14739e-06 (* 0.0272727 = 3.12925e-08 loss)
I0607 03:42:43.063140 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 7.74862e-07 (* 0.0272727 = 2.11326e-08 loss)
I0607 03:42:43.063154 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 1.2517e-06 (* 0.0272727 = 3.41373e-08 loss)
I0607 03:42:43.063168 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.8
I0607 03:42:43.063180 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 1
I0607 03:42:43.063192 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 1
I0607 03:42:43.063205 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.875
I0607 03:42:43.063216 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.625
I0607 03:42:43.063228 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.75
I0607 03:42:43.063241 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.875
I0607 03:42:43.063253 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.875
I0607 03:42:43.063266 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.875
I0607 03:42:43.063277 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.875
I0607 03:42:43.063289 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.875
I0607 03:42:43.063302 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.875
I0607 03:42:43.063314 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.875
I0607 03:42:43.063326 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.875
I0607 03:42:43.063338 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 1
I0607 03:42:43.063350 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 03:42:43.063361 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 03:42:43.063374 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 03:42:43.063385 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 03:42:43.063397 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 03:42:43.063410 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 03:42:43.063421 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 03:42:43.063432 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 03:42:43.063444 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.943182
I0607 03:42:43.063457 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.955556
I0607 03:42:43.063470 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 0.660004 (* 0.3 = 0.198001 loss)
I0607 03:42:43.063484 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.185454 (* 0.3 = 0.0556363 loss)
I0607 03:42:43.063510 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.123995 (* 0.0272727 = 0.00338168 loss)
I0607 03:42:43.063526 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.223502 (* 0.0272727 = 0.00609552 loss)
I0607 03:42:43.063541 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.517634 (* 0.0272727 = 0.0141173 loss)
I0607 03:42:43.063551 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.04073 (* 0.0272727 = 0.0283835 loss)
I0607 03:42:43.063561 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 0.616082 (* 0.0272727 = 0.0168022 loss)
I0607 03:42:43.063576 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 0.356736 (* 0.0272727 = 0.00972917 loss)
I0607 03:42:43.063591 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.433833 (* 0.0272727 = 0.0118318 loss)
I0607 03:42:43.063606 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.314876 (* 0.0272727 = 0.00858753 loss)
I0607 03:42:43.063621 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.245934 (* 0.0272727 = 0.00670728 loss)
I0607 03:42:43.063634 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.53333 (* 0.0272727 = 0.0145454 loss)
I0607 03:42:43.063648 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.317585 (* 0.0272727 = 0.00866142 loss)
I0607 03:42:43.063666 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.520266 (* 0.0272727 = 0.0141891 loss)
I0607 03:42:43.063681 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.711969 (* 0.0272727 = 0.0194173 loss)
I0607 03:42:43.063696 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.0015132 (* 0.0272727 = 4.12691e-05 loss)
I0607 03:42:43.063711 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.000477404 (* 0.0272727 = 1.30201e-05 loss)
I0607 03:42:43.063725 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 6.6812e-05 (* 0.0272727 = 1.82215e-06 loss)
I0607 03:42:43.063740 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 3.12292e-05 (* 0.0272727 = 8.51706e-07 loss)
I0607 03:42:43.063755 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 1.22344e-05 (* 0.0272727 = 3.33666e-07 loss)
I0607 03:42:43.063768 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 1.22193e-05 (* 0.0272727 = 3.33255e-07 loss)
I0607 03:42:43.063782 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 1.66894e-06 (* 0.0272727 = 4.55165e-08 loss)
I0607 03:42:43.063797 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 1.89246e-06 (* 0.0272727 = 5.16125e-08 loss)
I0607 03:42:43.063813 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 1.83285e-06 (* 0.0272727 = 4.99868e-08 loss)
I0607 03:42:43.063827 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.955556
I0607 03:42:43.063839 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 03:42:43.063851 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 1
I0607 03:42:43.063863 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 1
I0607 03:42:43.063874 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 1
I0607 03:42:43.063886 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.75
I0607 03:42:43.063899 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 1
I0607 03:42:43.063910 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.875
I0607 03:42:43.063923 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 1
I0607 03:42:43.063935 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 1
I0607 03:42:43.063947 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 1
I0607 03:42:43.063958 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.875
I0607 03:42:43.063971 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.875
I0607 03:42:43.063982 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 0.875
I0607 03:42:43.064005 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 03:42:43.064018 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 03:42:43.064030 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 03:42:43.064043 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 03:42:43.064054 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 03:42:43.064066 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 03:42:43.064079 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 03:42:43.064090 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 03:42:43.064102 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 03:42:43.064115 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.988636
I0607 03:42:43.064126 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 1
I0607 03:42:43.064141 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.233762 (* 1 = 0.233762 loss)
I0607 03:42:43.064155 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.0703463 (* 1 = 0.0703463 loss)
I0607 03:42:43.064170 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.0606227 (* 0.0909091 = 0.00551115 loss)
I0607 03:42:43.064184 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.0806238 (* 0.0909091 = 0.00732944 loss)
I0607 03:42:43.064198 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.169432 (* 0.0909091 = 0.0154029 loss)
I0607 03:42:43.064213 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.193458 (* 0.0909091 = 0.0175871 loss)
I0607 03:42:43.064226 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.703398 (* 0.0909091 = 0.0639453 loss)
I0607 03:42:43.064240 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.129466 (* 0.0909091 = 0.0117696 loss)
I0607 03:42:43.064254 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.236898 (* 0.0909091 = 0.0215361 loss)
I0607 03:42:43.064268 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.0801093 (* 0.0909091 = 0.00728267 loss)
I0607 03:42:43.064285 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.0385744 (* 0.0909091 = 0.00350676 loss)
I0607 03:42:43.064299 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.109673 (* 0.0909091 = 0.00997028 loss)
I0607 03:42:43.064314 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.166908 (* 0.0909091 = 0.0151734 loss)
I0607 03:42:43.064328 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.162826 (* 0.0909091 = 0.0148024 loss)
I0607 03:42:43.064342 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.280857 (* 0.0909091 = 0.0255324 loss)
I0607 03:42:43.064357 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.0591711 (* 0.0909091 = 0.00537919 loss)
I0607 03:42:43.064371 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.0171327 (* 0.0909091 = 0.00155752 loss)
I0607 03:42:43.064385 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.00718326 (* 0.0909091 = 0.000653024 loss)
I0607 03:42:43.064399 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.00269746 (* 0.0909091 = 0.000245224 loss)
I0607 03:42:43.064414 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.00041239 (* 0.0909091 = 3.749e-05 loss)
I0607 03:42:43.064429 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.000385824 (* 0.0909091 = 3.50749e-05 loss)
I0607 03:42:43.064443 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.000123109 (* 0.0909091 = 1.11917e-05 loss)
I0607 03:42:43.064457 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.000102215 (* 0.0909091 = 9.29226e-06 loss)
I0607 03:42:43.064471 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 0.000316315 (* 0.0909091 = 2.87559e-05 loss)
I0607 03:42:43.064493 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.875
I0607 03:42:43.064507 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.75
I0607 03:42:43.064519 32403 solver.cpp:245] Train net output #149: total_confidence = 0.620061
I0607 03:42:43.064532 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.525647
I0607 03:42:43.064544 32403 sgd_solver.cpp:106] Iteration 12000, lr = 0.001
I0607 03:43:39.120489 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.8243 > 30) by scale factor 0.886937
I0607 03:43:49.947932 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 51.3456 > 30) by scale factor 0.584276
I0607 03:45:14.234942 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 54.6013 > 30) by scale factor 0.549437
I0607 03:45:43.634768 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.7551 > 30) by scale factor 0.794595
I0607 03:45:59.140056 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.2734 > 30) by scale factor 0.929557
I0607 03:48:01.266482 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.708 > 30) by scale factor 0.889997
I0607 03:49:09.724398 32403 solver.cpp:229] Iteration 12500, loss = 3.9954
I0607 03:49:09.724498 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.72093
I0607 03:49:09.724519 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.75
I0607 03:49:09.724534 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.875
I0607 03:49:09.724547 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.75
I0607 03:49:09.724560 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.5
I0607 03:49:09.724573 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.375
I0607 03:49:09.724586 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.5
I0607 03:49:09.724599 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.875
I0607 03:49:09.724611 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 1
I0607 03:49:09.724625 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 1
I0607 03:49:09.724637 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 1
I0607 03:49:09.724650 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 1
I0607 03:49:09.724663 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 1
I0607 03:49:09.724674 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 1
I0607 03:49:09.724689 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 1
I0607 03:49:09.724702 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 03:49:09.724714 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 03:49:09.724726 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 03:49:09.724738 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 03:49:09.724750 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 03:49:09.724762 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 03:49:09.724774 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 03:49:09.724786 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 03:49:09.724798 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.914773
I0607 03:49:09.724812 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.860465
I0607 03:49:09.724828 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.12762 (* 0.3 = 0.338286 loss)
I0607 03:49:09.724843 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.320744 (* 0.3 = 0.0962231 loss)
I0607 03:49:09.724858 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.470833 (* 0.0272727 = 0.0128409 loss)
I0607 03:49:09.724872 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.05443 (* 0.0272727 = 0.0287573 loss)
I0607 03:49:09.724886 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.52844 (* 0.0272727 = 0.0416847 loss)
I0607 03:49:09.724900 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.22467 (* 0.0272727 = 0.0334002 loss)
I0607 03:49:09.724915 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 2.2197 (* 0.0272727 = 0.0605374 loss)
I0607 03:49:09.724928 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 2.31573 (* 0.0272727 = 0.0631562 loss)
I0607 03:49:09.724942 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 0.507505 (* 0.0272727 = 0.0138411 loss)
I0607 03:49:09.724957 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.0020101 (* 0.0272727 = 5.48208e-05 loss)
I0607 03:49:09.724972 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.000551456 (* 0.0272727 = 1.50397e-05 loss)
I0607 03:49:09.724987 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.000119368 (* 0.0272727 = 3.2555e-06 loss)
I0607 03:49:09.725002 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 1.8091e-05 (* 0.0272727 = 4.9339e-07 loss)
I0607 03:49:09.725015 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 4.69393e-06 (* 0.0272727 = 1.28016e-07 loss)
I0607 03:49:09.725049 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 5.2155e-06 (* 0.0272727 = 1.42241e-07 loss)
I0607 03:49:09.725065 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 2.07127e-06 (* 0.0272727 = 5.64893e-08 loss)
I0607 03:49:09.725080 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 1.1921e-06 (* 0.0272727 = 3.25117e-08 loss)
I0607 03:49:09.725095 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 9.53677e-07 (* 0.0272727 = 2.60094e-08 loss)
I0607 03:49:09.725108 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 7.49551e-06 (* 0.0272727 = 2.04423e-07 loss)
I0607 03:49:09.725137 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 1.54973e-06 (* 0.0272727 = 4.22653e-08 loss)
I0607 03:49:09.725153 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 4.60452e-06 (* 0.0272727 = 1.25578e-07 loss)
I0607 03:49:09.725168 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 1.80305e-06 (* 0.0272727 = 4.91741e-08 loss)
I0607 03:49:09.725183 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 1.71364e-06 (* 0.0272727 = 4.67357e-08 loss)
I0607 03:49:09.725196 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 3.11437e-06 (* 0.0272727 = 8.49375e-08 loss)
I0607 03:49:09.725208 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.883721
I0607 03:49:09.725221 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 1
I0607 03:49:09.725236 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 1
I0607 03:49:09.725250 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.875
I0607 03:49:09.725261 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.75
I0607 03:49:09.725273 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.625
I0607 03:49:09.725286 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.75
I0607 03:49:09.725297 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 1
I0607 03:49:09.725309 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 1
I0607 03:49:09.725322 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 1
I0607 03:49:09.725333 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 1
I0607 03:49:09.725344 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 1
I0607 03:49:09.725356 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 1
I0607 03:49:09.725368 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 1
I0607 03:49:09.725380 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 1
I0607 03:49:09.725391 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 03:49:09.725404 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 03:49:09.725415 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 03:49:09.725427 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 03:49:09.725438 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 03:49:09.725458 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 03:49:09.725476 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 03:49:09.725488 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 03:49:09.725504 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.960227
I0607 03:49:09.725520 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.930233
I0607 03:49:09.725535 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 0.560792 (* 0.3 = 0.168238 loss)
I0607 03:49:09.725550 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.16667 (* 0.3 = 0.0500011 loss)
I0607 03:49:09.725574 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.141447 (* 0.0272727 = 0.00385764 loss)
I0607 03:49:09.725589 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.342189 (* 0.0272727 = 0.00933242 loss)
I0607 03:49:09.725617 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.630403 (* 0.0272727 = 0.0171928 loss)
I0607 03:49:09.725633 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 0.436022 (* 0.0272727 = 0.0118915 loss)
I0607 03:49:09.725659 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.32795 (* 0.0272727 = 0.0362169 loss)
I0607 03:49:09.725687 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.9489 (* 0.0272727 = 0.0531519 loss)
I0607 03:49:09.725708 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.0533776 (* 0.0272727 = 0.00145575 loss)
I0607 03:49:09.725723 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.00433728 (* 0.0272727 = 0.00011829 loss)
I0607 03:49:09.725740 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.000669877 (* 0.0272727 = 1.82694e-05 loss)
I0607 03:49:09.725755 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.000196975 (* 0.0272727 = 5.37204e-06 loss)
I0607 03:49:09.725769 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.000108137 (* 0.0272727 = 2.94918e-06 loss)
I0607 03:49:09.725785 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 7.61937e-05 (* 0.0272727 = 2.07801e-06 loss)
I0607 03:49:09.725798 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 3.73387e-05 (* 0.0272727 = 1.01833e-06 loss)
I0607 03:49:09.725813 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 3.85598e-05 (* 0.0272727 = 1.05163e-06 loss)
I0607 03:49:09.725827 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 3.89324e-05 (* 0.0272727 = 1.06179e-06 loss)
I0607 03:49:09.725841 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 6.43742e-06 (* 0.0272727 = 1.75566e-07 loss)
I0607 03:49:09.725855 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 9.00048e-06 (* 0.0272727 = 2.45468e-07 loss)
I0607 03:49:09.725870 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 4.99194e-06 (* 0.0272727 = 1.36144e-07 loss)
I0607 03:49:09.725884 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 5.30487e-06 (* 0.0272727 = 1.44678e-07 loss)
I0607 03:49:09.725898 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 1.06398e-05 (* 0.0272727 = 2.90175e-07 loss)
I0607 03:49:09.725914 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 3.23357e-06 (* 0.0272727 = 8.81883e-08 loss)
I0607 03:49:09.725927 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 1.13998e-05 (* 0.0272727 = 3.10904e-07 loss)
I0607 03:49:09.725939 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.930233
I0607 03:49:09.725952 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 03:49:09.725965 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.875
I0607 03:49:09.725976 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.875
I0607 03:49:09.725987 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.875
I0607 03:49:09.725999 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.75
I0607 03:49:09.726012 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.75
I0607 03:49:09.726024 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 1
I0607 03:49:09.726037 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 1
I0607 03:49:09.726047 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 1
I0607 03:49:09.726059 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 1
I0607 03:49:09.726071 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 1
I0607 03:49:09.726083 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 1
I0607 03:49:09.726094 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 1
I0607 03:49:09.726106 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 03:49:09.726119 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 03:49:09.726138 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 03:49:09.726153 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 03:49:09.726166 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 03:49:09.726177 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 03:49:09.726189 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 03:49:09.726200 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 03:49:09.726212 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 03:49:09.726224 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.971591
I0607 03:49:09.726236 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.953488
I0607 03:49:09.726251 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.478853 (* 1 = 0.478853 loss)
I0607 03:49:09.726265 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.151007 (* 1 = 0.151007 loss)
I0607 03:49:09.726284 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.0829418 (* 0.0909091 = 0.00754017 loss)
I0607 03:49:09.726300 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.654182 (* 0.0909091 = 0.0594711 loss)
I0607 03:49:09.726313 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.177011 (* 0.0909091 = 0.0160919 loss)
I0607 03:49:09.726327 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.307974 (* 0.0909091 = 0.0279976 loss)
I0607 03:49:09.726342 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.666137 (* 0.0909091 = 0.0605579 loss)
I0607 03:49:09.726356 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 1.80644 (* 0.0909091 = 0.164222 loss)
I0607 03:49:09.726371 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.0395494 (* 0.0909091 = 0.0035954 loss)
I0607 03:49:09.726384 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.00347505 (* 0.0909091 = 0.000315914 loss)
I0607 03:49:09.726399 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.000534254 (* 0.0909091 = 4.85686e-05 loss)
I0607 03:49:09.726413 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.000462979 (* 0.0909091 = 4.2089e-05 loss)
I0607 03:49:09.726428 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.000323156 (* 0.0909091 = 2.93778e-05 loss)
I0607 03:49:09.726441 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.000123713 (* 0.0909091 = 1.12466e-05 loss)
I0607 03:49:09.726455 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 7.00424e-05 (* 0.0909091 = 6.36749e-06 loss)
I0607 03:49:09.726469 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 6.23004e-05 (* 0.0909091 = 5.66367e-06 loss)
I0607 03:49:09.726485 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 5.8003e-05 (* 0.0909091 = 5.273e-06 loss)
I0607 03:49:09.726498 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 3.86708e-05 (* 0.0909091 = 3.51553e-06 loss)
I0607 03:49:09.726513 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 4.6837e-05 (* 0.0909091 = 4.2579e-06 loss)
I0607 03:49:09.726526 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 5.07204e-05 (* 0.0909091 = 4.61094e-06 loss)
I0607 03:49:09.726541 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 6.56166e-05 (* 0.0909091 = 5.96515e-06 loss)
I0607 03:49:09.726555 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 6.53342e-05 (* 0.0909091 = 5.93947e-06 loss)
I0607 03:49:09.726569 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 2.40217e-05 (* 0.0909091 = 2.18379e-06 loss)
I0607 03:49:09.726584 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 4.90011e-05 (* 0.0909091 = 4.45465e-06 loss)
I0607 03:49:09.726596 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.625
I0607 03:49:09.726608 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.625
I0607 03:49:09.726631 32403 solver.cpp:245] Train net output #149: total_confidence = 0.557147
I0607 03:49:09.726644 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.533664
I0607 03:49:09.726658 32403 sgd_solver.cpp:106] Iteration 12500, lr = 0.001
I0607 03:49:22.487227 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 46.3586 > 30) by scale factor 0.647129
I0607 03:49:40.256403 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 41.7189 > 30) by scale factor 0.719098
I0607 03:50:18.118533 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 48.1854 > 30) by scale factor 0.622596
I0607 03:52:48.912870 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 36.8487 > 30) by scale factor 0.814141
I0607 03:53:34.492010 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.6767 > 30) by scale factor 0.890822
I0607 03:55:29.646878 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 35.7788 > 30) by scale factor 0.838486
I0607 03:55:36.241634 32403 solver.cpp:229] Iteration 13000, loss = 4.12927
I0607 03:55:36.241729 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.571429
I0607 03:55:36.241749 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.875
I0607 03:55:36.241762 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.625
I0607 03:55:36.241776 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.75
I0607 03:55:36.241789 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.25
I0607 03:55:36.241802 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.5
I0607 03:55:36.241816 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.5
I0607 03:55:36.241828 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.75
I0607 03:55:36.241842 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.875
I0607 03:55:36.241854 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.875
I0607 03:55:36.241868 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.875
I0607 03:55:36.241880 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.875
I0607 03:55:36.241894 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 03:55:36.241906 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 1
I0607 03:55:36.241919 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 1
I0607 03:55:36.241937 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 03:55:36.241950 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 03:55:36.241961 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 03:55:36.241973 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 03:55:36.241986 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 03:55:36.242005 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 03:55:36.242017 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 03:55:36.242029 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 03:55:36.242041 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.852273
I0607 03:55:36.242061 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.767857
I0607 03:55:36.242079 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.54151 (* 0.3 = 0.462452 loss)
I0607 03:55:36.242092 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.528611 (* 0.3 = 0.158583 loss)
I0607 03:55:36.242107 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.441935 (* 0.0272727 = 0.0120528 loss)
I0607 03:55:36.242130 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.37839 (* 0.0272727 = 0.0375924 loss)
I0607 03:55:36.242143 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.55398 (* 0.0272727 = 0.0423812 loss)
I0607 03:55:36.242157 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 2.63928 (* 0.0272727 = 0.0719802 loss)
I0607 03:55:36.242172 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.56799 (* 0.0272727 = 0.0427634 loss)
I0607 03:55:36.242187 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.82035 (* 0.0272727 = 0.0496458 loss)
I0607 03:55:36.242200 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 1.42663 (* 0.0272727 = 0.0389081 loss)
I0607 03:55:36.242214 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.608024 (* 0.0272727 = 0.0165825 loss)
I0607 03:55:36.242229 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.332356 (* 0.0272727 = 0.00906426 loss)
I0607 03:55:36.242244 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.380737 (* 0.0272727 = 0.0103837 loss)
I0607 03:55:36.242259 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.347337 (* 0.0272727 = 0.00947282 loss)
I0607 03:55:36.242274 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.486715 (* 0.0272727 = 0.0132741 loss)
I0607 03:55:36.242336 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.0371184 (* 0.0272727 = 0.00101232 loss)
I0607 03:55:36.242352 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.0108393 (* 0.0272727 = 0.000295617 loss)
I0607 03:55:36.242367 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.00714024 (* 0.0272727 = 0.000194734 loss)
I0607 03:55:36.242383 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.00353189 (* 0.0272727 = 9.63243e-05 loss)
I0607 03:55:36.242396 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.00450422 (* 0.0272727 = 0.000122842 loss)
I0607 03:55:36.242411 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.00176141 (* 0.0272727 = 4.80385e-05 loss)
I0607 03:55:36.242425 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.00112762 (* 0.0272727 = 3.07534e-05 loss)
I0607 03:55:36.242440 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.000202655 (* 0.0272727 = 5.52696e-06 loss)
I0607 03:55:36.242455 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.000263056 (* 0.0272727 = 7.17426e-06 loss)
I0607 03:55:36.242470 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 9.12794e-05 (* 0.0272727 = 2.48944e-06 loss)
I0607 03:55:36.242481 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.732143
I0607 03:55:36.242494 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 1
I0607 03:55:36.242507 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.875
I0607 03:55:36.242519 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.75
I0607 03:55:36.242532 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.875
I0607 03:55:36.242543 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.375
I0607 03:55:36.242555 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.625
I0607 03:55:36.242568 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.875
I0607 03:55:36.242579 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.875
I0607 03:55:36.242593 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.875
I0607 03:55:36.242604 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.875
I0607 03:55:36.242616 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.875
I0607 03:55:36.242629 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.875
I0607 03:55:36.242640 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 1
I0607 03:55:36.242652 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 1
I0607 03:55:36.242665 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 03:55:36.242676 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 03:55:36.242687 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 03:55:36.242699 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 03:55:36.242712 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 03:55:36.242723 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 03:55:36.242734 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 03:55:36.242749 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 03:55:36.242761 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.903409
I0607 03:55:36.242774 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.875
I0607 03:55:36.242787 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 0.977134 (* 0.3 = 0.29314 loss)
I0607 03:55:36.242802 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.356138 (* 0.3 = 0.106841 loss)
I0607 03:55:36.242816 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.178463 (* 0.0272727 = 0.00486717 loss)
I0607 03:55:36.242843 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.502942 (* 0.0272727 = 0.0137166 loss)
I0607 03:55:36.242858 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.935963 (* 0.0272727 = 0.0255263 loss)
I0607 03:55:36.242872 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 0.832604 (* 0.0272727 = 0.0227074 loss)
I0607 03:55:36.242887 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.78572 (* 0.0272727 = 0.0487016 loss)
I0607 03:55:36.242900 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.06279 (* 0.0272727 = 0.0289852 loss)
I0607 03:55:36.242914 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.598736 (* 0.0272727 = 0.0163292 loss)
I0607 03:55:36.242928 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.607374 (* 0.0272727 = 0.0165648 loss)
I0607 03:55:36.242943 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.367241 (* 0.0272727 = 0.0100157 loss)
I0607 03:55:36.242957 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.340473 (* 0.0272727 = 0.00928564 loss)
I0607 03:55:36.242971 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.298606 (* 0.0272727 = 0.00814379 loss)
I0607 03:55:36.242986 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.378625 (* 0.0272727 = 0.0103261 loss)
I0607 03:55:36.242997 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.116374 (* 0.0272727 = 0.00317384 loss)
I0607 03:55:36.243007 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.113793 (* 0.0272727 = 0.00310345 loss)
I0607 03:55:36.243018 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.0516716 (* 0.0272727 = 0.00140922 loss)
I0607 03:55:36.243032 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.0329352 (* 0.0272727 = 0.000898234 loss)
I0607 03:55:36.243047 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.0416261 (* 0.0272727 = 0.00113526 loss)
I0607 03:55:36.243062 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.0220741 (* 0.0272727 = 0.000602022 loss)
I0607 03:55:36.243077 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.00749921 (* 0.0272727 = 0.000204524 loss)
I0607 03:55:36.243090 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.00356608 (* 0.0272727 = 9.72567e-05 loss)
I0607 03:55:36.243104 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.00372135 (* 0.0272727 = 0.000101491 loss)
I0607 03:55:36.243119 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 0.000954503 (* 0.0272727 = 2.60319e-05 loss)
I0607 03:55:36.243131 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.928571
I0607 03:55:36.243144 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 03:55:36.243156 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 1
I0607 03:55:36.243168 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.875
I0607 03:55:36.243180 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 1
I0607 03:55:36.243192 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.875
I0607 03:55:36.243204 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 1
I0607 03:55:36.243216 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 1
I0607 03:55:36.243228 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.875
I0607 03:55:36.243240 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 1
I0607 03:55:36.243252 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.875
I0607 03:55:36.243264 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 1
I0607 03:55:36.243276 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.875
I0607 03:55:36.243288 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 1
I0607 03:55:36.243300 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 03:55:36.243312 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 03:55:36.243333 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 03:55:36.243347 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 03:55:36.243362 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 03:55:36.243376 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 03:55:36.243387 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 03:55:36.243399 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 03:55:36.243412 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 03:55:36.243422 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.971591
I0607 03:55:36.243434 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.964286
I0607 03:55:36.243448 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.433263 (* 1 = 0.433263 loss)
I0607 03:55:36.243463 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.150936 (* 1 = 0.150936 loss)
I0607 03:55:36.243477 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.105195 (* 0.0909091 = 0.00956323 loss)
I0607 03:55:36.243491 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.0865536 (* 0.0909091 = 0.00786851 loss)
I0607 03:55:36.243506 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.272502 (* 0.0909091 = 0.0247729 loss)
I0607 03:55:36.243521 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.144116 (* 0.0909091 = 0.0131014 loss)
I0607 03:55:36.243535 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.883391 (* 0.0909091 = 0.0803083 loss)
I0607 03:55:36.243551 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.422912 (* 0.0909091 = 0.0384466 loss)
I0607 03:55:36.243564 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.197024 (* 0.0909091 = 0.0179113 loss)
I0607 03:55:36.243578 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.258298 (* 0.0909091 = 0.0234817 loss)
I0607 03:55:36.243592 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.170357 (* 0.0909091 = 0.015487 loss)
I0607 03:55:36.243607 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.227637 (* 0.0909091 = 0.0206942 loss)
I0607 03:55:36.243621 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.162899 (* 0.0909091 = 0.014809 loss)
I0607 03:55:36.243635 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.330857 (* 0.0909091 = 0.0300779 loss)
I0607 03:55:36.243649 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.0356576 (* 0.0909091 = 0.0032416 loss)
I0607 03:55:36.243664 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.0115198 (* 0.0909091 = 0.00104725 loss)
I0607 03:55:36.243679 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.00113314 (* 0.0909091 = 0.000103012 loss)
I0607 03:55:36.243692 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.000451986 (* 0.0909091 = 4.10896e-05 loss)
I0607 03:55:36.243703 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.000288475 (* 0.0909091 = 2.6225e-05 loss)
I0607 03:55:36.243713 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.000200372 (* 0.0909091 = 1.82156e-05 loss)
I0607 03:55:36.243731 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.000130486 (* 0.0909091 = 1.18623e-05 loss)
I0607 03:55:36.243744 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 4.18679e-05 (* 0.0909091 = 3.80617e-06 loss)
I0607 03:55:36.243758 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 1.36944e-05 (* 0.0909091 = 1.24494e-06 loss)
I0607 03:55:36.243772 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 2.02809e-05 (* 0.0909091 = 1.84372e-06 loss)
I0607 03:55:36.243785 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.75
I0607 03:55:36.243810 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.75
I0607 03:55:36.243824 32403 solver.cpp:245] Train net output #149: total_confidence = 0.313743
I0607 03:55:36.243836 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.255935
I0607 03:55:36.243849 32403 sgd_solver.cpp:106] Iteration 13000, lr = 0.001
I0607 03:56:08.260315 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.8438 > 30) by scale factor 0.972644
I0607 03:56:19.070046 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 42.7779 > 30) by scale factor 0.701296
I0607 03:56:31.429285 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.6221 > 30) by scale factor 0.979686
I0607 03:56:32.976495 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 43.7221 > 30) by scale factor 0.686152
I0607 03:56:42.232585 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.186 > 30) by scale factor 0.932083
I0607 03:57:32.444532 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.3195 > 30) by scale factor 0.874139
I0607 03:58:13.393390 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.4106 > 30) by scale factor 0.925624
I0607 03:58:48.149972 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.504 > 30) by scale factor 0.922964
I0607 03:59:04.368937 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.6806 > 30) by scale factor 0.865038
I0607 03:59:21.345829 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 36.6752 > 30) by scale factor 0.817992
I0607 03:59:57.655057 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.2864 > 30) by scale factor 0.929184
I0607 04:00:23.157732 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.3167 > 30) by scale factor 0.957954
I0607 04:00:31.655952 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 41.347 > 30) by scale factor 0.725567
I0607 04:00:54.837792 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 44.8921 > 30) by scale factor 0.668269
I0607 04:01:14.164829 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.841 > 30) by scale factor 0.913491
I0607 04:02:02.076453 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 36.4562 > 30) by scale factor 0.822906
I0607 04:02:02.490689 32403 solver.cpp:229] Iteration 13500, loss = 4.03221
I0607 04:02:02.490748 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.353846
I0607 04:02:02.490767 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.375
I0607 04:02:02.490782 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.375
I0607 04:02:02.490795 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.375
I0607 04:02:02.490809 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.5
I0607 04:02:02.490823 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.625
I0607 04:02:02.490835 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.5
I0607 04:02:02.490847 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.75
I0607 04:02:02.490860 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.75
I0607 04:02:02.490874 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.625
I0607 04:02:02.490886 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.75
I0607 04:02:02.490900 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.625
I0607 04:02:02.490911 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.75
I0607 04:02:02.490924 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.75
I0607 04:02:02.490936 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.75
I0607 04:02:02.490949 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 0.875
I0607 04:02:02.490962 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 0.875
I0607 04:02:02.490973 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 04:02:02.490985 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 04:02:02.490998 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 04:02:02.491009 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 04:02:02.491021 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 04:02:02.491034 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 04:02:02.491045 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.744318
I0607 04:02:02.491057 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.661538
I0607 04:02:02.491075 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 2.26555 (* 0.3 = 0.679667 loss)
I0607 04:02:02.491089 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.897599 (* 0.3 = 0.26928 loss)
I0607 04:02:02.491103 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 2.28073 (* 0.0272727 = 0.0622016 loss)
I0607 04:02:02.491118 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 2.08977 (* 0.0272727 = 0.0569936 loss)
I0607 04:02:02.491132 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 2.10167 (* 0.0272727 = 0.0573182 loss)
I0607 04:02:02.491147 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.77381 (* 0.0272727 = 0.0483766 loss)
I0607 04:02:02.491160 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.68116 (* 0.0272727 = 0.0458498 loss)
I0607 04:02:02.491174 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.45041 (* 0.0272727 = 0.0395567 loss)
I0607 04:02:02.491189 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 0.884497 (* 0.0272727 = 0.0241227 loss)
I0607 04:02:02.491202 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.679666 (* 0.0272727 = 0.0185364 loss)
I0607 04:02:02.491217 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 1.1639 (* 0.0272727 = 0.0317427 loss)
I0607 04:02:02.491231 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 1.21453 (* 0.0272727 = 0.0331236 loss)
I0607 04:02:02.491245 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 1.21182 (* 0.0272727 = 0.0330496 loss)
I0607 04:02:02.491260 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 1.0236 (* 0.0272727 = 0.0279164 loss)
I0607 04:02:02.491307 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.607603 (* 0.0272727 = 0.016571 loss)
I0607 04:02:02.491327 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 1.46957 (* 0.0272727 = 0.0400792 loss)
I0607 04:02:02.491340 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.63251 (* 0.0272727 = 0.0172503 loss)
I0607 04:02:02.491354 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.686869 (* 0.0272727 = 0.0187328 loss)
I0607 04:02:02.491369 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.0124344 (* 0.0272727 = 0.000339121 loss)
I0607 04:02:02.491384 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.00529918 (* 0.0272727 = 0.000144523 loss)
I0607 04:02:02.491397 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.00289604 (* 0.0272727 = 7.8983e-05 loss)
I0607 04:02:02.491412 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.000837096 (* 0.0272727 = 2.28299e-05 loss)
I0607 04:02:02.491426 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.000363983 (* 0.0272727 = 9.9268e-06 loss)
I0607 04:02:02.491441 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.000170621 (* 0.0272727 = 4.65331e-06 loss)
I0607 04:02:02.491453 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.553846
I0607 04:02:02.491466 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.625
I0607 04:02:02.491479 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 1
I0607 04:02:02.491492 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.75
I0607 04:02:02.491503 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.875
I0607 04:02:02.491515 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.75
I0607 04:02:02.491528 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.625
I0607 04:02:02.491539 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.75
I0607 04:02:02.491551 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.75
I0607 04:02:02.491564 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.625
I0607 04:02:02.491575 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.75
I0607 04:02:02.491587 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.625
I0607 04:02:02.491600 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.75
I0607 04:02:02.491611 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.75
I0607 04:02:02.491623 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.75
I0607 04:02:02.491634 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 0.875
I0607 04:02:02.491647 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 0.875
I0607 04:02:02.491659 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 04:02:02.491672 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 04:02:02.491683 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 04:02:02.491695 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 04:02:02.491708 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 04:02:02.491719 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 04:02:02.491730 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.823864
I0607 04:02:02.491742 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.784615
I0607 04:02:02.491756 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.59832 (* 0.3 = 0.479497 loss)
I0607 04:02:02.491773 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.653782 (* 0.3 = 0.196135 loss)
I0607 04:02:02.491787 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.851627 (* 0.0272727 = 0.0232262 loss)
I0607 04:02:02.491813 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.46745 (* 0.0272727 = 0.0127486 loss)
I0607 04:02:02.491828 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 1.38654 (* 0.0272727 = 0.0378146 loss)
I0607 04:02:02.491842 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.17237 (* 0.0272727 = 0.0319736 loss)
I0607 04:02:02.491857 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.59716 (* 0.0272727 = 0.0435588 loss)
I0607 04:02:02.491870 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.17121 (* 0.0272727 = 0.031942 loss)
I0607 04:02:02.491884 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.610542 (* 0.0272727 = 0.0166511 loss)
I0607 04:02:02.491899 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.596049 (* 0.0272727 = 0.0162559 loss)
I0607 04:02:02.491912 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.902485 (* 0.0272727 = 0.0246132 loss)
I0607 04:02:02.491926 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.909375 (* 0.0272727 = 0.0248011 loss)
I0607 04:02:02.491940 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 1.00836 (* 0.0272727 = 0.0275006 loss)
I0607 04:02:02.491955 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.90213 (* 0.0272727 = 0.0246036 loss)
I0607 04:02:02.491968 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.658327 (* 0.0272727 = 0.0179544 loss)
I0607 04:02:02.491981 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 1.14471 (* 0.0272727 = 0.0312195 loss)
I0607 04:02:02.491997 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.338085 (* 0.0272727 = 0.00922049 loss)
I0607 04:02:02.492010 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.475475 (* 0.0272727 = 0.0129675 loss)
I0607 04:02:02.492024 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.0605431 (* 0.0272727 = 0.00165118 loss)
I0607 04:02:02.492039 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.0276455 (* 0.0272727 = 0.000753967 loss)
I0607 04:02:02.492053 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.00574542 (* 0.0272727 = 0.000156693 loss)
I0607 04:02:02.492068 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.00428689 (* 0.0272727 = 0.000116915 loss)
I0607 04:02:02.492082 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.000795135 (* 0.0272727 = 2.16855e-05 loss)
I0607 04:02:02.492096 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 0.000487005 (* 0.0272727 = 1.3282e-05 loss)
I0607 04:02:02.492108 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.723077
I0607 04:02:02.492121 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 0.625
I0607 04:02:02.492133 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.75
I0607 04:02:02.492146 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.75
I0607 04:02:02.492158 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.625
I0607 04:02:02.492171 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.75
I0607 04:02:02.492182 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.875
I0607 04:02:02.492193 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.75
I0607 04:02:02.492205 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.75
I0607 04:02:02.492218 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.875
I0607 04:02:02.492229 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 1
I0607 04:02:02.492241 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.625
I0607 04:02:02.492254 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.75
I0607 04:02:02.492265 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 0.75
I0607 04:02:02.492277 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 0.875
I0607 04:02:02.492288 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 0.875
I0607 04:02:02.492311 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 0.875
I0607 04:02:02.492321 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 04:02:02.492328 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 04:02:02.492336 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 04:02:02.492348 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 04:02:02.492359 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 04:02:02.492375 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 04:02:02.492388 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.880682
I0607 04:02:02.492400 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.892308
I0607 04:02:02.492414 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.980184 (* 1 = 0.980184 loss)
I0607 04:02:02.492429 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.428164 (* 1 = 0.428164 loss)
I0607 04:02:02.492444 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.90397 (* 0.0909091 = 0.0821791 loss)
I0607 04:02:02.492457 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.781389 (* 0.0909091 = 0.0710354 loss)
I0607 04:02:02.492471 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.882969 (* 0.0909091 = 0.0802699 loss)
I0607 04:02:02.492486 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.876146 (* 0.0909091 = 0.0796496 loss)
I0607 04:02:02.492501 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 1.02103 (* 0.0909091 = 0.0928213 loss)
I0607 04:02:02.492514 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.295689 (* 0.0909091 = 0.0268809 loss)
I0607 04:02:02.492528 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.723872 (* 0.0909091 = 0.0658065 loss)
I0607 04:02:02.492542 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.686388 (* 0.0909091 = 0.0623989 loss)
I0607 04:02:02.492557 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.477324 (* 0.0909091 = 0.0433931 loss)
I0607 04:02:02.492571 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.399176 (* 0.0909091 = 0.0362887 loss)
I0607 04:02:02.492585 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.729082 (* 0.0909091 = 0.0662801 loss)
I0607 04:02:02.492599 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.664864 (* 0.0909091 = 0.0604422 loss)
I0607 04:02:02.492614 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.583951 (* 0.0909091 = 0.0530865 loss)
I0607 04:02:02.492627 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.955163 (* 0.0909091 = 0.086833 loss)
I0607 04:02:02.492641 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.245186 (* 0.0909091 = 0.0222896 loss)
I0607 04:02:02.492655 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.156771 (* 0.0909091 = 0.0142519 loss)
I0607 04:02:02.492671 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.0258723 (* 0.0909091 = 0.00235203 loss)
I0607 04:02:02.492684 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.013045 (* 0.0909091 = 0.00118591 loss)
I0607 04:02:02.492698 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.00528556 (* 0.0909091 = 0.000480506 loss)
I0607 04:02:02.492712 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.00131962 (* 0.0909091 = 0.000119965 loss)
I0607 04:02:02.492727 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.00029686 (* 0.0909091 = 2.69873e-05 loss)
I0607 04:02:02.492741 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 0.000275256 (* 0.0909091 = 2.50232e-05 loss)
I0607 04:02:02.492753 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.375
I0607 04:02:02.492765 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.25
I0607 04:02:02.492786 32403 solver.cpp:245] Train net output #149: total_confidence = 0.218276
I0607 04:02:02.492799 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.247448
I0607 04:02:02.492815 32403 sgd_solver.cpp:106] Iteration 13500, lr = 0.001
I0607 04:02:20.638219 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.3849 > 30) by scale factor 0.898611
I0607 04:04:10.296164 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.3307 > 30) by scale factor 0.957528
I0607 04:07:30.480701 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.5001 > 30) by scale factor 0.923075
I0607 04:08:28.865253 32403 solver.cpp:229] Iteration 14000, loss = 4.06264
I0607 04:08:28.865396 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.411765
I0607 04:08:28.865417 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.625
I0607 04:08:28.865432 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.375
I0607 04:08:28.865444 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.25
I0607 04:08:28.865458 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.625
I0607 04:08:28.865470 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.5
I0607 04:08:28.865483 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.5
I0607 04:08:28.865496 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.5
I0607 04:08:28.865509 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.75
I0607 04:08:28.865521 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 1
I0607 04:08:28.865535 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 1
I0607 04:08:28.865547 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 1
I0607 04:08:28.865559 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 1
I0607 04:08:28.865571 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 1
I0607 04:08:28.865584 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 1
I0607 04:08:28.865597 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 04:08:28.865608 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 04:08:28.865620 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 04:08:28.865633 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 04:08:28.865644 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 04:08:28.865656 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 04:08:28.865669 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 04:08:28.865680 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 04:08:28.865692 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.823864
I0607 04:08:28.865705 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.764706
I0607 04:08:28.865721 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.74601 (* 0.3 = 0.523803 loss)
I0607 04:08:28.865736 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.532453 (* 0.3 = 0.159736 loss)
I0607 04:08:28.865751 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.775742 (* 0.0272727 = 0.0211566 loss)
I0607 04:08:28.865766 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.66138 (* 0.0272727 = 0.0453104 loss)
I0607 04:08:28.865779 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 2.06543 (* 0.0272727 = 0.0563299 loss)
I0607 04:08:28.865793 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.67374 (* 0.0272727 = 0.0456473 loss)
I0607 04:08:28.865808 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.85002 (* 0.0272727 = 0.0504552 loss)
I0607 04:08:28.865821 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 2.06292 (* 0.0272727 = 0.0562614 loss)
I0607 04:08:28.865835 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 2.07297 (* 0.0272727 = 0.0565356 loss)
I0607 04:08:28.865850 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 1.0525 (* 0.0272727 = 0.0287045 loss)
I0607 04:08:28.865864 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.00699963 (* 0.0272727 = 0.000190899 loss)
I0607 04:08:28.865882 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.00152422 (* 0.0272727 = 4.15697e-05 loss)
I0607 04:08:28.865897 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.000827597 (* 0.0272727 = 2.25708e-05 loss)
I0607 04:08:28.865911 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.0001357 (* 0.0272727 = 3.70092e-06 loss)
I0607 04:08:28.865947 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 8.94202e-05 (* 0.0272727 = 2.43873e-06 loss)
I0607 04:08:28.865962 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 4.40896e-05 (* 0.0272727 = 1.20244e-06 loss)
I0607 04:08:28.865978 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 6.89977e-05 (* 0.0272727 = 1.88176e-06 loss)
I0607 04:08:28.865993 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 4.07816e-05 (* 0.0272727 = 1.11223e-06 loss)
I0607 04:08:28.866006 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 4.39655e-05 (* 0.0272727 = 1.19906e-06 loss)
I0607 04:08:28.866021 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 8.65603e-05 (* 0.0272727 = 2.36074e-06 loss)
I0607 04:08:28.866035 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 4.81248e-05 (* 0.0272727 = 1.3125e-06 loss)
I0607 04:08:28.866050 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 5.99765e-05 (* 0.0272727 = 1.63572e-06 loss)
I0607 04:08:28.866063 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 4.27139e-05 (* 0.0272727 = 1.16492e-06 loss)
I0607 04:08:28.866078 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 4.75138e-05 (* 0.0272727 = 1.29583e-06 loss)
I0607 04:08:28.866091 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.627451
I0607 04:08:28.866103 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.875
I0607 04:08:28.866116 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 1
I0607 04:08:28.866127 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.75
I0607 04:08:28.866139 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.625
I0607 04:08:28.866152 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.375
I0607 04:08:28.866163 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.75
I0607 04:08:28.866175 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.625
I0607 04:08:28.866189 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.875
I0607 04:08:28.866200 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 1
I0607 04:08:28.866212 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 1
I0607 04:08:28.866224 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 1
I0607 04:08:28.866235 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 1
I0607 04:08:28.866247 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 1
I0607 04:08:28.866260 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 1
I0607 04:08:28.866271 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 04:08:28.866282 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 04:08:28.866294 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 04:08:28.866307 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 04:08:28.866317 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 04:08:28.866329 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 04:08:28.866340 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 04:08:28.866353 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 04:08:28.866364 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.886364
I0607 04:08:28.866375 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.921569
I0607 04:08:28.866390 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.04853 (* 0.3 = 0.314559 loss)
I0607 04:08:28.866406 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.330948 (* 0.3 = 0.0992843 loss)
I0607 04:08:28.866422 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.500679 (* 0.0272727 = 0.0136549 loss)
I0607 04:08:28.866437 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.284772 (* 0.0272727 = 0.00776652 loss)
I0607 04:08:28.866463 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.869274 (* 0.0272727 = 0.0237075 loss)
I0607 04:08:28.866479 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.65499 (* 0.0272727 = 0.0451361 loss)
I0607 04:08:28.866493 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.52813 (* 0.0272727 = 0.0416764 loss)
I0607 04:08:28.866508 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.24091 (* 0.0272727 = 0.0338429 loss)
I0607 04:08:28.866523 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.778489 (* 0.0272727 = 0.0212315 loss)
I0607 04:08:28.866538 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.325406 (* 0.0272727 = 0.00887472 loss)
I0607 04:08:28.866551 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.020551 (* 0.0272727 = 0.000560482 loss)
I0607 04:08:28.866565 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.0067015 (* 0.0272727 = 0.000182768 loss)
I0607 04:08:28.866580 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.00126219 (* 0.0272727 = 3.44232e-05 loss)
I0607 04:08:28.866595 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.000339934 (* 0.0272727 = 9.27092e-06 loss)
I0607 04:08:28.866608 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.000164379 (* 0.0272727 = 4.48307e-06 loss)
I0607 04:08:28.866622 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 5.22841e-05 (* 0.0272727 = 1.42593e-06 loss)
I0607 04:08:28.866636 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 3.79253e-05 (* 0.0272727 = 1.03433e-06 loss)
I0607 04:08:28.866652 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 2.35894e-05 (* 0.0272727 = 6.43346e-07 loss)
I0607 04:08:28.866665 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 1.01926e-05 (* 0.0272727 = 2.7798e-07 loss)
I0607 04:08:28.866679 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 3.41238e-06 (* 0.0272727 = 9.3065e-08 loss)
I0607 04:08:28.866694 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 7.27193e-06 (* 0.0272727 = 1.98325e-07 loss)
I0607 04:08:28.866708 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 1.66894e-06 (* 0.0272727 = 4.55164e-08 loss)
I0607 04:08:28.866724 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 1.06398e-05 (* 0.0272727 = 2.90175e-07 loss)
I0607 04:08:28.866737 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 4.27668e-06 (* 0.0272727 = 1.16637e-07 loss)
I0607 04:08:28.866750 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.803922
I0607 04:08:28.866762 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 04:08:28.866775 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.75
I0607 04:08:28.866786 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.875
I0607 04:08:28.866798 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.75
I0607 04:08:28.866811 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.875
I0607 04:08:28.866822 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.75
I0607 04:08:28.866834 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.875
I0607 04:08:28.866847 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.75
I0607 04:08:28.866859 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 1
I0607 04:08:28.866870 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 1
I0607 04:08:28.866883 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 1
I0607 04:08:28.866894 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 1
I0607 04:08:28.866905 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 1
I0607 04:08:28.866917 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 04:08:28.866931 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 04:08:28.866955 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 04:08:28.866967 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 04:08:28.866979 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 04:08:28.866991 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 04:08:28.867002 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 04:08:28.867014 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 04:08:28.867027 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 04:08:28.867038 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.943182
I0607 04:08:28.867051 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.901961
I0607 04:08:28.867065 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.652314 (* 1 = 0.652314 loss)
I0607 04:08:28.867079 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.201962 (* 1 = 0.201962 loss)
I0607 04:08:28.867094 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.119206 (* 0.0909091 = 0.0108369 loss)
I0607 04:08:28.867108 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.475471 (* 0.0909091 = 0.0432247 loss)
I0607 04:08:28.867122 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.45438 (* 0.0909091 = 0.0413073 loss)
I0607 04:08:28.867137 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.925041 (* 0.0909091 = 0.0840947 loss)
I0607 04:08:28.867151 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.369595 (* 0.0909091 = 0.0335996 loss)
I0607 04:08:28.867166 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.676224 (* 0.0909091 = 0.0614749 loss)
I0607 04:08:28.867179 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.662508 (* 0.0909091 = 0.060228 loss)
I0607 04:08:28.867194 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.688697 (* 0.0909091 = 0.0626088 loss)
I0607 04:08:28.867208 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.0135427 (* 0.0909091 = 0.00123116 loss)
I0607 04:08:28.867223 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.00379752 (* 0.0909091 = 0.00034523 loss)
I0607 04:08:28.867233 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.00166605 (* 0.0909091 = 0.000151459 loss)
I0607 04:08:28.867244 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.00114953 (* 0.0909091 = 0.000104503 loss)
I0607 04:08:28.867259 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.000456302 (* 0.0909091 = 4.1482e-05 loss)
I0607 04:08:28.867274 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.000679802 (* 0.0909091 = 6.18002e-05 loss)
I0607 04:08:28.867287 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.000305614 (* 0.0909091 = 2.77831e-05 loss)
I0607 04:08:28.867301 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.000239916 (* 0.0909091 = 2.18105e-05 loss)
I0607 04:08:28.867316 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.000342312 (* 0.0909091 = 3.11192e-05 loss)
I0607 04:08:28.867329 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.000246818 (* 0.0909091 = 2.2438e-05 loss)
I0607 04:08:28.867343 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.000484761 (* 0.0909091 = 4.40691e-05 loss)
I0607 04:08:28.867358 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.000517319 (* 0.0909091 = 4.7029e-05 loss)
I0607 04:08:28.867372 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.000305124 (* 0.0909091 = 2.77386e-05 loss)
I0607 04:08:28.867386 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 0.000280264 (* 0.0909091 = 2.54785e-05 loss)
I0607 04:08:28.867398 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.375
I0607 04:08:28.867410 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.25
I0607 04:08:28.867432 32403 solver.cpp:245] Train net output #149: total_confidence = 0.321337
I0607 04:08:28.867445 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.242437
I0607 04:08:28.867461 32403 sgd_solver.cpp:106] Iteration 14000, lr = 0.001
I0607 04:09:27.254922 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.1773 > 30) by scale factor 0.904234
I0607 04:10:46.139292 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.8244 > 30) by scale factor 0.973255
I0607 04:11:58.752140 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.6446 > 30) by scale factor 0.865936
I0607 04:12:38.165778 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 38.9174 > 30) by scale factor 0.770864
I0607 04:12:45.106410 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 41.5239 > 30) by scale factor 0.722476
I0607 04:12:49.758422 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 38.5139 > 30) by scale factor 0.77894
I0607 04:14:55.347031 32403 solver.cpp:229] Iteration 14500, loss = 4.1509
I0607 04:14:55.347120 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.534483
I0607 04:14:55.347139 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.5
I0607 04:14:55.347153 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.625
I0607 04:14:55.347167 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.375
I0607 04:14:55.347179 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.5
I0607 04:14:55.347193 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.5
I0607 04:14:55.347205 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.375
I0607 04:14:55.347218 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.5
I0607 04:14:55.347230 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.75
I0607 04:14:55.347244 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.875
I0607 04:14:55.347256 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.875
I0607 04:14:55.347270 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.875
I0607 04:14:55.347281 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 04:14:55.347295 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 1
I0607 04:14:55.347306 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 1
I0607 04:14:55.347318 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 04:14:55.347331 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 04:14:55.347342 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 04:14:55.347353 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 04:14:55.347365 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 04:14:55.347378 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 04:14:55.347389 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 04:14:55.347400 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 04:14:55.347412 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.846591
I0607 04:14:55.347425 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.775862
I0607 04:14:55.347441 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.52766 (* 0.3 = 0.458297 loss)
I0607 04:14:55.347456 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.520847 (* 0.3 = 0.156254 loss)
I0607 04:14:55.347470 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 1.38354 (* 0.0272727 = 0.0377329 loss)
I0607 04:14:55.347484 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 0.845171 (* 0.0272727 = 0.0230501 loss)
I0607 04:14:55.347498 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.6766 (* 0.0272727 = 0.0457254 loss)
I0607 04:14:55.347512 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.3108 (* 0.0272727 = 0.0357492 loss)
I0607 04:14:55.347527 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.91587 (* 0.0272727 = 0.0522509 loss)
I0607 04:14:55.347540 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.73652 (* 0.0272727 = 0.0473596 loss)
I0607 04:14:55.347553 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 1.55013 (* 0.0272727 = 0.0422764 loss)
I0607 04:14:55.347568 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 1.04345 (* 0.0272727 = 0.0284576 loss)
I0607 04:14:55.347582 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.274328 (* 0.0272727 = 0.00748167 loss)
I0607 04:14:55.347597 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.301011 (* 0.0272727 = 0.0082094 loss)
I0607 04:14:55.347611 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.474529 (* 0.0272727 = 0.0129417 loss)
I0607 04:14:55.347626 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.265493 (* 0.0272727 = 0.00724072 loss)
I0607 04:14:55.347659 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.127351 (* 0.0272727 = 0.00347321 loss)
I0607 04:14:55.347676 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.0310919 (* 0.0272727 = 0.000847961 loss)
I0607 04:14:55.347690 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.00461615 (* 0.0272727 = 0.000125895 loss)
I0607 04:14:55.347705 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.000715653 (* 0.0272727 = 1.95178e-05 loss)
I0607 04:14:55.347719 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.000484855 (* 0.0272727 = 1.32233e-05 loss)
I0607 04:14:55.347735 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 4.63696e-05 (* 0.0272727 = 1.26463e-06 loss)
I0607 04:14:55.347750 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 2.26215e-05 (* 0.0272727 = 6.16951e-07 loss)
I0607 04:14:55.347765 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 1.28601e-05 (* 0.0272727 = 3.5073e-07 loss)
I0607 04:14:55.347779 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 3.82963e-06 (* 0.0272727 = 1.04444e-07 loss)
I0607 04:14:55.347793 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 3.63591e-06 (* 0.0272727 = 9.91612e-08 loss)
I0607 04:14:55.347805 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.689655
I0607 04:14:55.347818 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.75
I0607 04:14:55.347831 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.875
I0607 04:14:55.347842 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.625
I0607 04:14:55.347854 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.5
I0607 04:14:55.347867 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.625
I0607 04:14:55.347879 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.5
I0607 04:14:55.347890 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.875
I0607 04:14:55.347903 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.75
I0607 04:14:55.347914 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.875
I0607 04:14:55.347926 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 1
I0607 04:14:55.347939 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.875
I0607 04:14:55.347949 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.875
I0607 04:14:55.347961 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 1
I0607 04:14:55.347973 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 1
I0607 04:14:55.347985 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 04:14:55.347996 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 04:14:55.348008 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 04:14:55.348019 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 04:14:55.348031 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 04:14:55.348042 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 04:14:55.348054 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 04:14:55.348067 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 04:14:55.348078 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.892045
I0607 04:14:55.348089 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.793103
I0607 04:14:55.348104 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.23387 (* 0.3 = 0.370162 loss)
I0607 04:14:55.348117 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.438636 (* 0.3 = 0.131591 loss)
I0607 04:14:55.348136 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 1.00884 (* 0.0272727 = 0.0275138 loss)
I0607 04:14:55.348150 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.389756 (* 0.0272727 = 0.0106297 loss)
I0607 04:14:55.348176 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.719348 (* 0.0272727 = 0.0196186 loss)
I0607 04:14:55.348191 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.55489 (* 0.0272727 = 0.0424061 loss)
I0607 04:14:55.348206 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.2507 (* 0.0272727 = 0.0341101 loss)
I0607 04:14:55.348219 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.3474 (* 0.0272727 = 0.0367472 loss)
I0607 04:14:55.348233 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.997229 (* 0.0272727 = 0.0271971 loss)
I0607 04:14:55.348243 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.633567 (* 0.0272727 = 0.0172791 loss)
I0607 04:14:55.348259 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.271034 (* 0.0272727 = 0.00739183 loss)
I0607 04:14:55.348273 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.13702 (* 0.0272727 = 0.0037369 loss)
I0607 04:14:55.348287 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.347256 (* 0.0272727 = 0.00947062 loss)
I0607 04:14:55.348302 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.378037 (* 0.0272727 = 0.0103101 loss)
I0607 04:14:55.348316 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.123565 (* 0.0272727 = 0.00336995 loss)
I0607 04:14:55.348330 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.0885759 (* 0.0272727 = 0.00241571 loss)
I0607 04:14:55.348345 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.028008 (* 0.0272727 = 0.000763854 loss)
I0607 04:14:55.348358 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.00390774 (* 0.0272727 = 0.000106575 loss)
I0607 04:14:55.348372 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.000391498 (* 0.0272727 = 1.06772e-05 loss)
I0607 04:14:55.348387 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.000131449 (* 0.0272727 = 3.58499e-06 loss)
I0607 04:14:55.348400 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 3.92255e-05 (* 0.0272727 = 1.06979e-06 loss)
I0607 04:14:55.348414 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 3.17398e-06 (* 0.0272727 = 8.6563e-08 loss)
I0607 04:14:55.348428 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 4.73864e-06 (* 0.0272727 = 1.29236e-07 loss)
I0607 04:14:55.348443 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 1.07289e-06 (* 0.0272727 = 2.92605e-08 loss)
I0607 04:14:55.348454 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.931035
I0607 04:14:55.348467 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 04:14:55.348479 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 1
I0607 04:14:55.348490 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 1
I0607 04:14:55.348502 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.875
I0607 04:14:55.348513 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.875
I0607 04:14:55.348526 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.75
I0607 04:14:55.348537 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.75
I0607 04:14:55.348549 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 1
I0607 04:14:55.348561 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 1
I0607 04:14:55.348572 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.875
I0607 04:14:55.348583 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 1
I0607 04:14:55.348595 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 1
I0607 04:14:55.348606 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 1
I0607 04:14:55.348618 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 04:14:55.348629 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 04:14:55.348641 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 04:14:55.348662 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 04:14:55.348675 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 04:14:55.348687 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 04:14:55.348700 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 04:14:55.348711 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 04:14:55.348721 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 04:14:55.348733 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.971591
I0607 04:14:55.348745 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.982759
I0607 04:14:55.348759 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.381085 (* 1 = 0.381085 loss)
I0607 04:14:55.348773 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.148031 (* 1 = 0.148031 loss)
I0607 04:14:55.348790 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.0990544 (* 0.0909091 = 0.00900494 loss)
I0607 04:14:55.348805 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.0350926 (* 0.0909091 = 0.00319024 loss)
I0607 04:14:55.348819 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.102384 (* 0.0909091 = 0.00930762 loss)
I0607 04:14:55.348834 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 1.01876 (* 0.0909091 = 0.0926144 loss)
I0607 04:14:55.348847 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.420858 (* 0.0909091 = 0.0382598 loss)
I0607 04:14:55.348861 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.313637 (* 0.0909091 = 0.0285125 loss)
I0607 04:14:55.348875 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.712187 (* 0.0909091 = 0.0647443 loss)
I0607 04:14:55.348889 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.166094 (* 0.0909091 = 0.0150995 loss)
I0607 04:14:55.348903 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.11183 (* 0.0909091 = 0.0101664 loss)
I0607 04:14:55.348917 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.209425 (* 0.0909091 = 0.0190387 loss)
I0607 04:14:55.348932 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.105095 (* 0.0909091 = 0.00955413 loss)
I0607 04:14:55.348945 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.105664 (* 0.0909091 = 0.00960579 loss)
I0607 04:14:55.348959 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.0568886 (* 0.0909091 = 0.00517169 loss)
I0607 04:14:55.348973 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.0898267 (* 0.0909091 = 0.00816607 loss)
I0607 04:14:55.348987 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.0107548 (* 0.0909091 = 0.000977708 loss)
I0607 04:14:55.349004 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.00288439 (* 0.0909091 = 0.000262217 loss)
I0607 04:14:55.349019 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.00078815 (* 0.0909091 = 7.165e-05 loss)
I0607 04:14:55.349032 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 9.73164e-05 (* 0.0909091 = 8.84695e-06 loss)
I0607 04:14:55.349046 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 4.31596e-05 (* 0.0909091 = 3.9236e-06 loss)
I0607 04:14:55.349061 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 2.1027e-05 (* 0.0909091 = 1.91154e-06 loss)
I0607 04:14:55.349074 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 7.00365e-06 (* 0.0909091 = 6.36696e-07 loss)
I0607 04:14:55.349088 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 7.01853e-06 (* 0.0909091 = 6.38048e-07 loss)
I0607 04:14:55.349100 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.625
I0607 04:14:55.349112 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.375
I0607 04:14:55.349149 32403 solver.cpp:245] Train net output #149: total_confidence = 0.316309
I0607 04:14:55.349164 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.343579
I0607 04:14:55.349179 32403 sgd_solver.cpp:106] Iteration 14500, lr = 0.001
I0607 04:15:15.037096 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 38.9904 > 30) by scale factor 0.76942
I0607 04:15:18.908826 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 39.9337 > 30) by scale factor 0.751246
I0607 04:15:30.491529 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 65.4419 > 30) by scale factor 0.458422
I0607 04:18:43.730162 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.0165 > 30) by scale factor 0.967228
I0607 04:18:58.418280 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.2336 > 30) by scale factor 0.805724
I0607 04:20:02.512840 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.3752 > 30) by scale factor 0.898872
I0607 04:20:17.982552 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.5313 > 30) by scale factor 0.982597
I0607 04:20:45.005982 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 57.7985 > 30) by scale factor 0.519045
I0607 04:20:56.596801 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.4712 > 30) by scale factor 0.953253
I0607 04:21:21.338153 32403 solver.cpp:338] Iteration 15000, Testing net (#0)
I0607 04:22:19.583746 32403 solver.cpp:393] Test loss: 2.61909
I0607 04:22:19.583878 32403 solver.cpp:406] Test net output #0: loss1/accuracy = 0.617407
I0607 04:22:19.583899 32403 solver.cpp:406] Test net output #1: loss1/accuracy01 = 0.798
I0607 04:22:19.583912 32403 solver.cpp:406] Test net output #2: loss1/accuracy02 = 0.669
I0607 04:22:19.583925 32403 solver.cpp:406] Test net output #3: loss1/accuracy03 = 0.546
I0607 04:22:19.583937 32403 solver.cpp:406] Test net output #4: loss1/accuracy04 = 0.476
I0607 04:22:19.583950 32403 solver.cpp:406] Test net output #5: loss1/accuracy05 = 0.528
I0607 04:22:19.583963 32403 solver.cpp:406] Test net output #6: loss1/accuracy06 = 0.721
I0607 04:22:19.583976 32403 solver.cpp:406] Test net output #7: loss1/accuracy07 = 0.848
I0607 04:22:19.583988 32403 solver.cpp:406] Test net output #8: loss1/accuracy08 = 0.922
I0607 04:22:19.584000 32403 solver.cpp:406] Test net output #9: loss1/accuracy09 = 0.971
I0607 04:22:19.584013 32403 solver.cpp:406] Test net output #10: loss1/accuracy10 = 0.986
I0607 04:22:19.584025 32403 solver.cpp:406] Test net output #11: loss1/accuracy11 = 0.995
I0607 04:22:19.584038 32403 solver.cpp:406] Test net output #12: loss1/accuracy12 = 0.999
I0607 04:22:19.584050 32403 solver.cpp:406] Test net output #13: loss1/accuracy13 = 1
I0607 04:22:19.584061 32403 solver.cpp:406] Test net output #14: loss1/accuracy14 = 1
I0607 04:22:19.584074 32403 solver.cpp:406] Test net output #15: loss1/accuracy15 = 1
I0607 04:22:19.584085 32403 solver.cpp:406] Test net output #16: loss1/accuracy16 = 1
I0607 04:22:19.584096 32403 solver.cpp:406] Test net output #17: loss1/accuracy17 = 1
I0607 04:22:19.584108 32403 solver.cpp:406] Test net output #18: loss1/accuracy18 = 1
I0607 04:22:19.584120 32403 solver.cpp:406] Test net output #19: loss1/accuracy19 = 1
I0607 04:22:19.584131 32403 solver.cpp:406] Test net output #20: loss1/accuracy20 = 1
I0607 04:22:19.584143 32403 solver.cpp:406] Test net output #21: loss1/accuracy21 = 1
I0607 04:22:19.584154 32403 solver.cpp:406] Test net output #22: loss1/accuracy22 = 1
I0607 04:22:19.584167 32403 solver.cpp:406] Test net output #23: loss1/accuracy_incl_empty = 0.89132
I0607 04:22:19.584177 32403 solver.cpp:406] Test net output #24: loss1/accuracy_top3 = 0.846871
I0607 04:22:19.584195 32403 solver.cpp:406] Test net output #25: loss1/cross_entropy_loss = 1.3764 (* 0.3 = 0.41292 loss)
I0607 04:22:19.584210 32403 solver.cpp:406] Test net output #26: loss1/cross_entropy_loss_incl_empty = 0.394295 (* 0.3 = 0.118289 loss)
I0607 04:22:19.584224 32403 solver.cpp:406] Test net output #27: loss1/loss01 = 0.905966 (* 0.0272727 = 0.0247082 loss)
I0607 04:22:19.584239 32403 solver.cpp:406] Test net output #28: loss1/loss02 = 1.32672 (* 0.0272727 = 0.0361832 loss)
I0607 04:22:19.584254 32403 solver.cpp:406] Test net output #29: loss1/loss03 = 1.64933 (* 0.0272727 = 0.0449817 loss)
I0607 04:22:19.584267 32403 solver.cpp:406] Test net output #30: loss1/loss04 = 1.8145 (* 0.0272727 = 0.0494865 loss)
I0607 04:22:19.584280 32403 solver.cpp:406] Test net output #31: loss1/loss05 = 1.55681 (* 0.0272727 = 0.0424585 loss)
I0607 04:22:19.584295 32403 solver.cpp:406] Test net output #32: loss1/loss06 = 0.971752 (* 0.0272727 = 0.0265023 loss)
I0607 04:22:19.584308 32403 solver.cpp:406] Test net output #33: loss1/loss07 = 0.517154 (* 0.0272727 = 0.0141042 loss)
I0607 04:22:19.584322 32403 solver.cpp:406] Test net output #34: loss1/loss08 = 0.279883 (* 0.0272727 = 0.00763317 loss)
I0607 04:22:19.584336 32403 solver.cpp:406] Test net output #35: loss1/loss09 = 0.140741 (* 0.0272727 = 0.00383839 loss)
I0607 04:22:19.584352 32403 solver.cpp:406] Test net output #36: loss1/loss10 = 0.0766893 (* 0.0272727 = 0.00209153 loss)
I0607 04:22:19.584365 32403 solver.cpp:406] Test net output #37: loss1/loss11 = 0.0271172 (* 0.0272727 = 0.00073956 loss)
I0607 04:22:19.584379 32403 solver.cpp:406] Test net output #38: loss1/loss12 = 0.0153915 (* 0.0272727 = 0.000419768 loss)
I0607 04:22:19.584393 32403 solver.cpp:406] Test net output #39: loss1/loss13 = 0.0105076 (* 0.0272727 = 0.000286571 loss)
I0607 04:22:19.584429 32403 solver.cpp:406] Test net output #40: loss1/loss14 = 0.00759854 (* 0.0272727 = 0.000207233 loss)
I0607 04:22:19.584444 32403 solver.cpp:406] Test net output #41: loss1/loss15 = 0.00531866 (* 0.0272727 = 0.000145054 loss)
I0607 04:22:19.584458 32403 solver.cpp:406] Test net output #42: loss1/loss16 = 0.00416319 (* 0.0272727 = 0.000113541 loss)
I0607 04:22:19.584472 32403 solver.cpp:406] Test net output #43: loss1/loss17 = 0.00383936 (* 0.0272727 = 0.00010471 loss)
I0607 04:22:19.584486 32403 solver.cpp:406] Test net output #44: loss1/loss18 = 0.0033709 (* 0.0272727 = 9.19337e-05 loss)
I0607 04:22:19.584501 32403 solver.cpp:406] Test net output #45: loss1/loss19 = 0.00270073 (* 0.0272727 = 7.36563e-05 loss)
I0607 04:22:19.584514 32403 solver.cpp:406] Test net output #46: loss1/loss20 = 0.00263094 (* 0.0272727 = 7.17529e-05 loss)
I0607 04:22:19.584528 32403 solver.cpp:406] Test net output #47: loss1/loss21 = 0.00205721 (* 0.0272727 = 5.61057e-05 loss)
I0607 04:22:19.584542 32403 solver.cpp:406] Test net output #48: loss1/loss22 = 0.00173485 (* 0.0272727 = 4.73141e-05 loss)
I0607 04:22:19.584555 32403 solver.cpp:406] Test net output #49: loss2/accuracy = 0.757283
I0607 04:22:19.584568 32403 solver.cpp:406] Test net output #50: loss2/accuracy01 = 0.876
I0607 04:22:19.584578 32403 solver.cpp:406] Test net output #51: loss2/accuracy02 = 0.818
I0607 04:22:19.584590 32403 solver.cpp:406] Test net output #52: loss2/accuracy03 = 0.769
I0607 04:22:19.584602 32403 solver.cpp:406] Test net output #53: loss2/accuracy04 = 0.645
I0607 04:22:19.584614 32403 solver.cpp:406] Test net output #54: loss2/accuracy05 = 0.666
I0607 04:22:19.584625 32403 solver.cpp:406] Test net output #55: loss2/accuracy06 = 0.781
I0607 04:22:19.584636 32403 solver.cpp:406] Test net output #56: loss2/accuracy07 = 0.88
I0607 04:22:19.584647 32403 solver.cpp:406] Test net output #57: loss2/accuracy08 = 0.934
I0607 04:22:19.584659 32403 solver.cpp:406] Test net output #58: loss2/accuracy09 = 0.971
I0607 04:22:19.584671 32403 solver.cpp:406] Test net output #59: loss2/accuracy10 = 0.985
I0607 04:22:19.584682 32403 solver.cpp:406] Test net output #60: loss2/accuracy11 = 0.995
I0607 04:22:19.584694 32403 solver.cpp:406] Test net output #61: loss2/accuracy12 = 0.999
I0607 04:22:19.584705 32403 solver.cpp:406] Test net output #62: loss2/accuracy13 = 1
I0607 04:22:19.584717 32403 solver.cpp:406] Test net output #63: loss2/accuracy14 = 1
I0607 04:22:19.584728 32403 solver.cpp:406] Test net output #64: loss2/accuracy15 = 1
I0607 04:22:19.584739 32403 solver.cpp:406] Test net output #65: loss2/accuracy16 = 1
I0607 04:22:19.584750 32403 solver.cpp:406] Test net output #66: loss2/accuracy17 = 1
I0607 04:22:19.584761 32403 solver.cpp:406] Test net output #67: loss2/accuracy18 = 1
I0607 04:22:19.584774 32403 solver.cpp:406] Test net output #68: loss2/accuracy19 = 1
I0607 04:22:19.584784 32403 solver.cpp:406] Test net output #69: loss2/accuracy20 = 1
I0607 04:22:19.584795 32403 solver.cpp:406] Test net output #70: loss2/accuracy21 = 1
I0607 04:22:19.584807 32403 solver.cpp:406] Test net output #71: loss2/accuracy22 = 1
I0607 04:22:19.584818 32403 solver.cpp:406] Test net output #72: loss2/accuracy_incl_empty = 0.929183
I0607 04:22:19.584830 32403 solver.cpp:406] Test net output #73: loss2/accuracy_top3 = 0.910503
I0607 04:22:19.584843 32403 solver.cpp:406] Test net output #74: loss2/cross_entropy_loss = 0.927458 (* 0.3 = 0.278237 loss)
I0607 04:22:19.584857 32403 solver.cpp:406] Test net output #75: loss2/cross_entropy_loss_incl_empty = 0.272441 (* 0.3 = 0.0817323 loss)
I0607 04:22:19.584872 32403 solver.cpp:406] Test net output #76: loss2/loss01 = 0.606145 (* 0.0272727 = 0.0165312 loss)
I0607 04:22:19.584890 32403 solver.cpp:406] Test net output #77: loss2/loss02 = 0.847146 (* 0.0272727 = 0.023104 loss)
I0607 04:22:19.584916 32403 solver.cpp:406] Test net output #78: loss2/loss03 = 0.984596 (* 0.0272727 = 0.0268526 loss)
I0607 04:22:19.584931 32403 solver.cpp:406] Test net output #79: loss2/loss04 = 1.22084 (* 0.0272727 = 0.0332957 loss)
I0607 04:22:19.584944 32403 solver.cpp:406] Test net output #80: loss2/loss05 = 1.1079 (* 0.0272727 = 0.0302155 loss)
I0607 04:22:19.584959 32403 solver.cpp:406] Test net output #81: loss2/loss06 = 0.739746 (* 0.0272727 = 0.0201749 loss)
I0607 04:22:19.584971 32403 solver.cpp:406] Test net output #82: loss2/loss07 = 0.415918 (* 0.0272727 = 0.0113432 loss)
I0607 04:22:19.584985 32403 solver.cpp:406] Test net output #83: loss2/loss08 = 0.22559 (* 0.0272727 = 0.00615245 loss)
I0607 04:22:19.585000 32403 solver.cpp:406] Test net output #84: loss2/loss09 = 0.127392 (* 0.0272727 = 0.00347434 loss)
I0607 04:22:19.585013 32403 solver.cpp:406] Test net output #85: loss2/loss10 = 0.0668859 (* 0.0272727 = 0.00182416 loss)
I0607 04:22:19.585027 32403 solver.cpp:406] Test net output #86: loss2/loss11 = 0.021998 (* 0.0272727 = 0.000599946 loss)
I0607 04:22:19.585041 32403 solver.cpp:406] Test net output #87: loss2/loss12 = 0.00967372 (* 0.0272727 = 0.000263829 loss)
I0607 04:22:19.585055 32403 solver.cpp:406] Test net output #88: loss2/loss13 = 0.00591736 (* 0.0272727 = 0.000161382 loss)
I0607 04:22:19.585068 32403 solver.cpp:406] Test net output #89: loss2/loss14 = 0.00374147 (* 0.0272727 = 0.00010204 loss)
I0607 04:22:19.585083 32403 solver.cpp:406] Test net output #90: loss2/loss15 = 0.00276511 (* 0.0272727 = 7.54122e-05 loss)
I0607 04:22:19.585096 32403 solver.cpp:406] Test net output #91: loss2/loss16 = 0.0023234 (* 0.0272727 = 6.33655e-05 loss)
I0607 04:22:19.585110 32403 solver.cpp:406] Test net output #92: loss2/loss17 = 0.00211216 (* 0.0272727 = 5.76043e-05 loss)
I0607 04:22:19.585139 32403 solver.cpp:406] Test net output #93: loss2/loss18 = 0.00214532 (* 0.0272727 = 5.85087e-05 loss)
I0607 04:22:19.585155 32403 solver.cpp:406] Test net output #94: loss2/loss19 = 0.00163834 (* 0.0272727 = 4.46819e-05 loss)
I0607 04:22:19.585170 32403 solver.cpp:406] Test net output #95: loss2/loss20 = 0.00177118 (* 0.0272727 = 4.83048e-05 loss)
I0607 04:22:19.585183 32403 solver.cpp:406] Test net output #96: loss2/loss21 = 0.00172956 (* 0.0272727 = 4.71697e-05 loss)
I0607 04:22:19.585197 32403 solver.cpp:406] Test net output #97: loss2/loss22 = 0.00162586 (* 0.0272727 = 4.43416e-05 loss)
I0607 04:22:19.585209 32403 solver.cpp:406] Test net output #98: loss3/accuracy = 0.843572
I0607 04:22:19.585222 32403 solver.cpp:406] Test net output #99: loss3/accuracy01 = 0.891
I0607 04:22:19.585233 32403 solver.cpp:406] Test net output #100: loss3/accuracy02 = 0.862
I0607 04:22:19.585244 32403 solver.cpp:406] Test net output #101: loss3/accuracy03 = 0.862
I0607 04:22:19.585256 32403 solver.cpp:406] Test net output #102: loss3/accuracy04 = 0.835
I0607 04:22:19.585268 32403 solver.cpp:406] Test net output #103: loss3/accuracy05 = 0.829
I0607 04:22:19.585279 32403 solver.cpp:406] Test net output #104: loss3/accuracy06 = 0.862
I0607 04:22:19.585290 32403 solver.cpp:406] Test net output #105: loss3/accuracy07 = 0.905
I0607 04:22:19.585302 32403 solver.cpp:406] Test net output #106: loss3/accuracy08 = 0.945
I0607 04:22:19.585314 32403 solver.cpp:406] Test net output #107: loss3/accuracy09 = 0.969
I0607 04:22:19.585325 32403 solver.cpp:406] Test net output #108: loss3/accuracy10 = 0.982
I0607 04:22:19.585336 32403 solver.cpp:406] Test net output #109: loss3/accuracy11 = 0.993
I0607 04:22:19.585347 32403 solver.cpp:406] Test net output #110: loss3/accuracy12 = 0.998
I0607 04:22:19.585360 32403 solver.cpp:406] Test net output #111: loss3/accuracy13 = 0.999
I0607 04:22:19.585371 32403 solver.cpp:406] Test net output #112: loss3/accuracy14 = 1
I0607 04:22:19.585381 32403 solver.cpp:406] Test net output #113: loss3/accuracy15 = 1
I0607 04:22:19.585392 32403 solver.cpp:406] Test net output #114: loss3/accuracy16 = 1
I0607 04:22:19.585415 32403 solver.cpp:406] Test net output #115: loss3/accuracy17 = 1
I0607 04:22:19.585425 32403 solver.cpp:406] Test net output #116: loss3/accuracy18 = 1
I0607 04:22:19.585433 32403 solver.cpp:406] Test net output #117: loss3/accuracy19 = 1
I0607 04:22:19.585444 32403 solver.cpp:406] Test net output #118: loss3/accuracy20 = 1
I0607 04:22:19.585456 32403 solver.cpp:406] Test net output #119: loss3/accuracy21 = 1
I0607 04:22:19.585467 32403 solver.cpp:406] Test net output #120: loss3/accuracy22 = 1
I0607 04:22:19.585479 32403 solver.cpp:406] Test net output #121: loss3/accuracy_incl_empty = 0.95141
I0607 04:22:19.585490 32403 solver.cpp:406] Test net output #122: loss3/accuracy_top3 = 0.928676
I0607 04:22:19.585505 32403 solver.cpp:406] Test net output #123: loss3/cross_entropy_loss = 0.673272 (* 1 = 0.673272 loss)
I0607 04:22:19.585517 32403 solver.cpp:406] Test net output #124: loss3/cross_entropy_loss_incl_empty = 0.211919 (* 1 = 0.211919 loss)
I0607 04:22:19.585532 32403 solver.cpp:406] Test net output #125: loss3/loss01 = 0.490842 (* 0.0909091 = 0.044622 loss)
I0607 04:22:19.585546 32403 solver.cpp:406] Test net output #126: loss3/loss02 = 0.65096 (* 0.0909091 = 0.0591782 loss)
I0607 04:22:19.585559 32403 solver.cpp:406] Test net output #127: loss3/loss03 = 0.64634 (* 0.0909091 = 0.0587582 loss)
I0607 04:22:19.585573 32403 solver.cpp:406] Test net output #128: loss3/loss04 = 0.718535 (* 0.0909091 = 0.0653214 loss)
I0607 04:22:19.585587 32403 solver.cpp:406] Test net output #129: loss3/loss05 = 0.701242 (* 0.0909091 = 0.0637493 loss)
I0607 04:22:19.585600 32403 solver.cpp:406] Test net output #130: loss3/loss06 = 0.531263 (* 0.0909091 = 0.0482966 loss)
I0607 04:22:19.585614 32403 solver.cpp:406] Test net output #131: loss3/loss07 = 0.358721 (* 0.0909091 = 0.032611 loss)
I0607 04:22:19.585628 32403 solver.cpp:406] Test net output #132: loss3/loss08 = 0.202992 (* 0.0909091 = 0.0184538 loss)
I0607 04:22:19.585641 32403 solver.cpp:406] Test net output #133: loss3/loss09 = 0.121545 (* 0.0909091 = 0.0110495 loss)
I0607 04:22:19.585656 32403 solver.cpp:406] Test net output #134: loss3/loss10 = 0.0752558 (* 0.0909091 = 0.00684144 loss)
I0607 04:22:19.585669 32403 solver.cpp:406] Test net output #135: loss3/loss11 = 0.0278378 (* 0.0909091 = 0.00253071 loss)
I0607 04:22:19.585683 32403 solver.cpp:406] Test net output #136: loss3/loss12 = 0.0118591 (* 0.0909091 = 0.0010781 loss)
I0607 04:22:19.585697 32403 solver.cpp:406] Test net output #137: loss3/loss13 = 0.00618331 (* 0.0909091 = 0.000562119 loss)
I0607 04:22:19.585711 32403 solver.cpp:406] Test net output #138: loss3/loss14 = 0.0032004 (* 0.0909091 = 0.000290945 loss)
I0607 04:22:19.585724 32403 solver.cpp:406] Test net output #139: loss3/loss15 = 0.00125389 (* 0.0909091 = 0.00011399 loss)
I0607 04:22:19.585738 32403 solver.cpp:406] Test net output #140: loss3/loss16 = 0.000784845 (* 0.0909091 = 7.13496e-05 loss)
I0607 04:22:19.585752 32403 solver.cpp:406] Test net output #141: loss3/loss17 = 0.000639803 (* 0.0909091 = 5.81639e-05 loss)
I0607 04:22:19.585767 32403 solver.cpp:406] Test net output #142: loss3/loss18 = 0.00060326 (* 0.0909091 = 5.48418e-05 loss)
I0607 04:22:19.585779 32403 solver.cpp:406] Test net output #143: loss3/loss19 = 0.000533426 (* 0.0909091 = 4.84932e-05 loss)
I0607 04:22:19.585793 32403 solver.cpp:406] Test net output #144: loss3/loss20 = 0.000578944 (* 0.0909091 = 5.26313e-05 loss)
I0607 04:22:19.585808 32403 solver.cpp:406] Test net output #145: loss3/loss21 = 0.000623105 (* 0.0909091 = 5.66459e-05 loss)
I0607 04:22:19.585820 32403 solver.cpp:406] Test net output #146: loss3/loss22 = 0.000484361 (* 0.0909091 = 4.40328e-05 loss)
I0607 04:22:19.585832 32403 solver.cpp:406] Test net output #147: total_accuracy = 0.582
I0607 04:22:19.585844 32403 solver.cpp:406] Test net output #148: total_accuracy_not_rec = 0.558
I0607 04:22:19.585855 32403 solver.cpp:406] Test net output #149: total_confidence = 0.50648
I0607 04:22:19.585876 32403 solver.cpp:406] Test net output #150: total_confidence_not_rec = 0.447678
I0607 04:22:19.585891 32403 solver.cpp:338] Iteration 15000, Testing net (#1)
I0607 04:23:18.013047 32403 solver.cpp:393] Test loss: 3.67323
I0607 04:23:18.013191 32403 solver.cpp:406] Test net output #0: loss1/accuracy = 0.550234
I0607 04:23:18.013213 32403 solver.cpp:406] Test net output #1: loss1/accuracy01 = 0.76
I0607 04:23:18.013226 32403 solver.cpp:406] Test net output #2: loss1/accuracy02 = 0.661
I0607 04:23:18.013239 32403 solver.cpp:406] Test net output #3: loss1/accuracy03 = 0.511
I0607 04:23:18.013252 32403 solver.cpp:406] Test net output #4: loss1/accuracy04 = 0.404
I0607 04:23:18.013265 32403 solver.cpp:406] Test net output #5: loss1/accuracy05 = 0.475
I0607 04:23:18.013278 32403 solver.cpp:406] Test net output #6: loss1/accuracy06 = 0.63
I0607 04:23:18.013290 32403 solver.cpp:406] Test net output #7: loss1/accuracy07 = 0.745
I0607 04:23:18.013303 32403 solver.cpp:406] Test net output #8: loss1/accuracy08 = 0.797
I0607 04:23:18.013315 32403 solver.cpp:406] Test net output #9: loss1/accuracy09 = 0.841
I0607 04:23:18.013329 32403 solver.cpp:406] Test net output #10: loss1/accuracy10 = 0.85
I0607 04:23:18.013340 32403 solver.cpp:406] Test net output #11: loss1/accuracy11 = 0.886
I0607 04:23:18.013352 32403 solver.cpp:406] Test net output #12: loss1/accuracy12 = 0.894
I0607 04:23:18.013365 32403 solver.cpp:406] Test net output #13: loss1/accuracy13 = 0.916
I0607 04:23:18.013377 32403 solver.cpp:406] Test net output #14: loss1/accuracy14 = 0.935
I0607 04:23:18.013388 32403 solver.cpp:406] Test net output #15: loss1/accuracy15 = 0.946
I0607 04:23:18.013401 32403 solver.cpp:406] Test net output #16: loss1/accuracy16 = 0.964
I0607 04:23:18.013412 32403 solver.cpp:406] Test net output #17: loss1/accuracy17 = 0.982
I0607 04:23:18.013424 32403 solver.cpp:406] Test net output #18: loss1/accuracy18 = 0.988
I0607 04:23:18.013437 32403 solver.cpp:406] Test net output #19: loss1/accuracy19 = 0.99
I0607 04:23:18.013448 32403 solver.cpp:406] Test net output #20: loss1/accuracy20 = 0.996
I0607 04:23:18.013460 32403 solver.cpp:406] Test net output #21: loss1/accuracy21 = 0.999
I0607 04:23:18.013473 32403 solver.cpp:406] Test net output #22: loss1/accuracy22 = 1
I0607 04:23:18.013484 32403 solver.cpp:406] Test net output #23: loss1/accuracy_incl_empty = 0.835001
I0607 04:23:18.013496 32403 solver.cpp:406] Test net output #24: loss1/accuracy_top3 = 0.784222
I0607 04:23:18.013512 32403 solver.cpp:406] Test net output #25: loss1/cross_entropy_loss = 1.64912 (* 0.3 = 0.494736 loss)
I0607 04:23:18.013527 32403 solver.cpp:406] Test net output #26: loss1/cross_entropy_loss_incl_empty = 0.603195 (* 0.3 = 0.180959 loss)
I0607 04:23:18.013541 32403 solver.cpp:406] Test net output #27: loss1/loss01 = 1.042 (* 0.0272727 = 0.0284182 loss)
I0607 04:23:18.013556 32403 solver.cpp:406] Test net output #28: loss1/loss02 = 1.41279 (* 0.0272727 = 0.0385307 loss)
I0607 04:23:18.013569 32403 solver.cpp:406] Test net output #29: loss1/loss03 = 1.77147 (* 0.0272727 = 0.0483127 loss)
I0607 04:23:18.013582 32403 solver.cpp:406] Test net output #30: loss1/loss04 = 2.00273 (* 0.0272727 = 0.0546198 loss)
I0607 04:23:18.013597 32403 solver.cpp:406] Test net output #31: loss1/loss05 = 1.76213 (* 0.0272727 = 0.0480581 loss)
I0607 04:23:18.013610 32403 solver.cpp:406] Test net output #32: loss1/loss06 = 1.26995 (* 0.0272727 = 0.034635 loss)
I0607 04:23:18.013624 32403 solver.cpp:406] Test net output #33: loss1/loss07 = 0.8915 (* 0.0272727 = 0.0243136 loss)
I0607 04:23:18.013638 32403 solver.cpp:406] Test net output #34: loss1/loss08 = 0.71752 (* 0.0272727 = 0.0195687 loss)
I0607 04:23:18.013653 32403 solver.cpp:406] Test net output #35: loss1/loss09 = 0.602275 (* 0.0272727 = 0.0164257 loss)
I0607 04:23:18.013666 32403 solver.cpp:406] Test net output #36: loss1/loss10 = 0.530019 (* 0.0272727 = 0.0144551 loss)
I0607 04:23:18.013680 32403 solver.cpp:406] Test net output #37: loss1/loss11 = 0.459956 (* 0.0272727 = 0.0125443 loss)
I0607 04:23:18.013694 32403 solver.cpp:406] Test net output #38: loss1/loss12 = 0.391285 (* 0.0272727 = 0.0106714 loss)
I0607 04:23:18.013708 32403 solver.cpp:406] Test net output #39: loss1/loss13 = 0.29239 (* 0.0272727 = 0.00797427 loss)
I0607 04:23:18.013742 32403 solver.cpp:406] Test net output #40: loss1/loss14 = 0.242621 (* 0.0272727 = 0.00661693 loss)
I0607 04:23:18.013757 32403 solver.cpp:406] Test net output #41: loss1/loss15 = 0.202843 (* 0.0272727 = 0.00553208 loss)
I0607 04:23:18.013772 32403 solver.cpp:406] Test net output #42: loss1/loss16 = 0.158078 (* 0.0272727 = 0.00431121 loss)
I0607 04:23:18.013787 32403 solver.cpp:406] Test net output #43: loss1/loss17 = 0.0960743 (* 0.0272727 = 0.00262021 loss)
I0607 04:23:18.013800 32403 solver.cpp:406] Test net output #44: loss1/loss18 = 0.0715609 (* 0.0272727 = 0.00195166 loss)
I0607 04:23:18.013814 32403 solver.cpp:406] Test net output #45: loss1/loss19 = 0.0615688 (* 0.0272727 = 0.00167915 loss)
I0607 04:23:18.013828 32403 solver.cpp:406] Test net output #46: loss1/loss20 = 0.0258351 (* 0.0272727 = 0.000704593 loss)
I0607 04:23:18.013842 32403 solver.cpp:406] Test net output #47: loss1/loss21 = 0.00916499 (* 0.0272727 = 0.000249954 loss)
I0607 04:23:18.013856 32403 solver.cpp:406] Test net output #48: loss1/loss22 = 0.00265868 (* 0.0272727 = 7.25096e-05 loss)
I0607 04:23:18.013870 32403 solver.cpp:406] Test net output #49: loss2/accuracy = 0.68755
I0607 04:23:18.013885 32403 solver.cpp:406] Test net output #50: loss2/accuracy01 = 0.853
I0607 04:23:18.013896 32403 solver.cpp:406] Test net output #51: loss2/accuracy02 = 0.822
I0607 04:23:18.013908 32403 solver.cpp:406] Test net output #52: loss2/accuracy03 = 0.721
I0607 04:23:18.013921 32403 solver.cpp:406] Test net output #53: loss2/accuracy04 = 0.609
I0607 04:23:18.013932 32403 solver.cpp:406] Test net output #54: loss2/accuracy05 = 0.603
I0607 04:23:18.013943 32403 solver.cpp:406] Test net output #55: loss2/accuracy06 = 0.725
I0607 04:23:18.013955 32403 solver.cpp:406] Test net output #56: loss2/accuracy07 = 0.775
I0607 04:23:18.013967 32403 solver.cpp:406] Test net output #57: loss2/accuracy08 = 0.832
I0607 04:23:18.013978 32403 solver.cpp:406] Test net output #58: loss2/accuracy09 = 0.853
I0607 04:23:18.013989 32403 solver.cpp:406] Test net output #59: loss2/accuracy10 = 0.864
I0607 04:23:18.014001 32403 solver.cpp:406] Test net output #60: loss2/accuracy11 = 0.89
I0607 04:23:18.014013 32403 solver.cpp:406] Test net output #61: loss2/accuracy12 = 0.904
I0607 04:23:18.014024 32403 solver.cpp:406] Test net output #62: loss2/accuracy13 = 0.922
I0607 04:23:18.014035 32403 solver.cpp:406] Test net output #63: loss2/accuracy14 = 0.936
I0607 04:23:18.014047 32403 solver.cpp:406] Test net output #64: loss2/accuracy15 = 0.946
I0607 04:23:18.014058 32403 solver.cpp:406] Test net output #65: loss2/accuracy16 = 0.965
I0607 04:23:18.014070 32403 solver.cpp:406] Test net output #66: loss2/accuracy17 = 0.982
I0607 04:23:18.014081 32403 solver.cpp:406] Test net output #67: loss2/accuracy18 = 0.988
I0607 04:23:18.014093 32403 solver.cpp:406] Test net output #68: loss2/accuracy19 = 0.99
I0607 04:23:18.014106 32403 solver.cpp:406] Test net output #69: loss2/accuracy20 = 0.996
I0607 04:23:18.014117 32403 solver.cpp:406] Test net output #70: loss2/accuracy21 = 0.999
I0607 04:23:18.014128 32403 solver.cpp:406] Test net output #71: loss2/accuracy22 = 1
I0607 04:23:18.014139 32403 solver.cpp:406] Test net output #72: loss2/accuracy_incl_empty = 0.876683
I0607 04:23:18.014152 32403 solver.cpp:406] Test net output #73: loss2/accuracy_top3 = 0.862383
I0607 04:23:18.014165 32403 solver.cpp:406] Test net output #74: loss2/cross_entropy_loss = 1.19203 (* 0.3 = 0.35761 loss)
I0607 04:23:18.014178 32403 solver.cpp:406] Test net output #75: loss2/cross_entropy_loss_incl_empty = 0.455266 (* 0.3 = 0.13658 loss)
I0607 04:23:18.014192 32403 solver.cpp:406] Test net output #76: loss2/loss01 = 0.731482 (* 0.0272727 = 0.0199495 loss)
I0607 04:23:18.014209 32403 solver.cpp:406] Test net output #77: loss2/loss02 = 0.885955 (* 0.0272727 = 0.0241624 loss)
I0607 04:23:18.014235 32403 solver.cpp:406] Test net output #78: loss2/loss03 = 1.13295 (* 0.0272727 = 0.0308985 loss)
I0607 04:23:18.014250 32403 solver.cpp:406] Test net output #79: loss2/loss04 = 1.36658 (* 0.0272727 = 0.0372703 loss)
I0607 04:23:18.014263 32403 solver.cpp:406] Test net output #80: loss2/loss05 = 1.29928 (* 0.0272727 = 0.035435 loss)
I0607 04:23:18.014277 32403 solver.cpp:406] Test net output #81: loss2/loss06 = 0.981942 (* 0.0272727 = 0.0267802 loss)
I0607 04:23:18.014291 32403 solver.cpp:406] Test net output #82: loss2/loss07 = 0.77049 (* 0.0272727 = 0.0210134 loss)
I0607 04:23:18.014304 32403 solver.cpp:406] Test net output #83: loss2/loss08 = 0.605692 (* 0.0272727 = 0.0165189 loss)
I0607 04:23:18.014318 32403 solver.cpp:406] Test net output #84: loss2/loss09 = 0.534721 (* 0.0272727 = 0.0145833 loss)
I0607 04:23:18.014331 32403 solver.cpp:406] Test net output #85: loss2/loss10 = 0.478628 (* 0.0272727 = 0.0130535 loss)
I0607 04:23:18.014345 32403 solver.cpp:406] Test net output #86: loss2/loss11 = 0.416854 (* 0.0272727 = 0.0113688 loss)
I0607 04:23:18.014355 32403 solver.cpp:406] Test net output #87: loss2/loss12 = 0.344333 (* 0.0272727 = 0.00939091 loss)
I0607 04:23:18.014369 32403 solver.cpp:406] Test net output #88: loss2/loss13 = 0.267747 (* 0.0272727 = 0.0073022 loss)
I0607 04:23:18.014384 32403 solver.cpp:406] Test net output #89: loss2/loss14 = 0.221276 (* 0.0272727 = 0.00603481 loss)
I0607 04:23:18.014397 32403 solver.cpp:406] Test net output #90: loss2/loss15 = 0.184965 (* 0.0272727 = 0.0050445 loss)
I0607 04:23:18.014411 32403 solver.cpp:406] Test net output #91: loss2/loss16 = 0.148138 (* 0.0272727 = 0.00404013 loss)
I0607 04:23:18.014425 32403 solver.cpp:406] Test net output #92: loss2/loss17 = 0.0843513 (* 0.0272727 = 0.00230049 loss)
I0607 04:23:18.014439 32403 solver.cpp:406] Test net output #93: loss2/loss18 = 0.062372 (* 0.0272727 = 0.00170106 loss)
I0607 04:23:18.014453 32403 solver.cpp:406] Test net output #94: loss2/loss19 = 0.0601969 (* 0.0272727 = 0.00164173 loss)
I0607 04:23:18.014467 32403 solver.cpp:406] Test net output #95: loss2/loss20 = 0.0237095 (* 0.0272727 = 0.000646622 loss)
I0607 04:23:18.014480 32403 solver.cpp:406] Test net output #96: loss2/loss21 = 0.00901162 (* 0.0272727 = 0.000245772 loss)
I0607 04:23:18.014494 32403 solver.cpp:406] Test net output #97: loss2/loss22 = 0.00251641 (* 0.0272727 = 6.86293e-05 loss)
I0607 04:23:18.014506 32403 solver.cpp:406] Test net output #98: loss3/accuracy = 0.797633
I0607 04:23:18.014518 32403 solver.cpp:406] Test net output #99: loss3/accuracy01 = 0.873
I0607 04:23:18.014530 32403 solver.cpp:406] Test net output #100: loss3/accuracy02 = 0.856
I0607 04:23:18.014542 32403 solver.cpp:406] Test net output #101: loss3/accuracy03 = 0.831
I0607 04:23:18.014554 32403 solver.cpp:406] Test net output #102: loss3/accuracy04 = 0.818
I0607 04:23:18.014565 32403 solver.cpp:406] Test net output #103: loss3/accuracy05 = 0.824
I0607 04:23:18.014576 32403 solver.cpp:406] Test net output #104: loss3/accuracy06 = 0.85
I0607 04:23:18.014588 32403 solver.cpp:406] Test net output #105: loss3/accuracy07 = 0.877
I0607 04:23:18.014600 32403 solver.cpp:406] Test net output #106: loss3/accuracy08 = 0.878
I0607 04:23:18.014611 32403 solver.cpp:406] Test net output #107: loss3/accuracy09 = 0.902
I0607 04:23:18.014622 32403 solver.cpp:406] Test net output #108: loss3/accuracy10 = 0.899
I0607 04:23:18.014634 32403 solver.cpp:406] Test net output #109: loss3/accuracy11 = 0.929
I0607 04:23:18.014645 32403 solver.cpp:406] Test net output #110: loss3/accuracy12 = 0.925
I0607 04:23:18.014657 32403 solver.cpp:406] Test net output #111: loss3/accuracy13 = 0.939
I0607 04:23:18.014668 32403 solver.cpp:406] Test net output #112: loss3/accuracy14 = 0.957
I0607 04:23:18.014679 32403 solver.cpp:406] Test net output #113: loss3/accuracy15 = 0.959
I0607 04:23:18.014691 32403 solver.cpp:406] Test net output #114: loss3/accuracy16 = 0.969
I0607 04:23:18.014713 32403 solver.cpp:406] Test net output #115: loss3/accuracy17 = 0.985
I0607 04:23:18.014725 32403 solver.cpp:406] Test net output #116: loss3/accuracy18 = 0.991
I0607 04:23:18.014737 32403 solver.cpp:406] Test net output #117: loss3/accuracy19 = 0.992
I0607 04:23:18.014749 32403 solver.cpp:406] Test net output #118: loss3/accuracy20 = 0.996
I0607 04:23:18.014760 32403 solver.cpp:406] Test net output #119: loss3/accuracy21 = 0.999
I0607 04:23:18.014771 32403 solver.cpp:406] Test net output #120: loss3/accuracy22 = 1
I0607 04:23:18.014782 32403 solver.cpp:406] Test net output #121: loss3/accuracy_incl_empty = 0.917637
I0607 04:23:18.014794 32403 solver.cpp:406] Test net output #122: loss3/accuracy_top3 = 0.907889
I0607 04:23:18.014807 32403 solver.cpp:406] Test net output #123: loss3/cross_entropy_loss = 0.840962 (* 1 = 0.840962 loss)
I0607 04:23:18.014822 32403 solver.cpp:406] Test net output #124: loss3/cross_entropy_loss_incl_empty = 0.327712 (* 1 = 0.327712 loss)
I0607 04:23:18.014835 32403 solver.cpp:406] Test net output #125: loss3/loss01 = 0.589312 (* 0.0909091 = 0.0535738 loss)
I0607 04:23:18.014848 32403 solver.cpp:406] Test net output #126: loss3/loss02 = 0.721299 (* 0.0909091 = 0.0655726 loss)
I0607 04:23:18.014863 32403 solver.cpp:406] Test net output #127: loss3/loss03 = 0.771836 (* 0.0909091 = 0.0701669 loss)
I0607 04:23:18.014875 32403 solver.cpp:406] Test net output #128: loss3/loss04 = 0.793414 (* 0.0909091 = 0.0721286 loss)
I0607 04:23:18.014889 32403 solver.cpp:406] Test net output #129: loss3/loss05 = 0.756236 (* 0.0909091 = 0.0687487 loss)
I0607 04:23:18.014902 32403 solver.cpp:406] Test net output #130: loss3/loss06 = 0.632204 (* 0.0909091 = 0.0574731 loss)
I0607 04:23:18.014916 32403 solver.cpp:406] Test net output #131: loss3/loss07 = 0.484519 (* 0.0909091 = 0.0440472 loss)
I0607 04:23:18.014932 32403 solver.cpp:406] Test net output #132: loss3/loss08 = 0.461307 (* 0.0909091 = 0.041937 loss)
I0607 04:23:18.014946 32403 solver.cpp:406] Test net output #133: loss3/loss09 = 0.376511 (* 0.0909091 = 0.0342283 loss)
I0607 04:23:18.014961 32403 solver.cpp:406] Test net output #134: loss3/loss10 = 0.356703 (* 0.0909091 = 0.0324276 loss)
I0607 04:23:18.014973 32403 solver.cpp:406] Test net output #135: loss3/loss11 = 0.29468 (* 0.0909091 = 0.0267891 loss)
I0607 04:23:18.014987 32403 solver.cpp:406] Test net output #136: loss3/loss12 = 0.260814 (* 0.0909091 = 0.0237104 loss)
I0607 04:23:18.015000 32403 solver.cpp:406] Test net output #137: loss3/loss13 = 0.212624 (* 0.0909091 = 0.0193294 loss)
I0607 04:23:18.015014 32403 solver.cpp:406] Test net output #138: loss3/loss14 = 0.153363 (* 0.0909091 = 0.0139421 loss)
I0607 04:23:18.015028 32403 solver.cpp:406] Test net output #139: loss3/loss15 = 0.140186 (* 0.0909091 = 0.0127442 loss)
I0607 04:23:18.015041 32403 solver.cpp:406] Test net output #140: loss3/loss16 = 0.105947 (* 0.0909091 = 0.00963154 loss)
I0607 04:23:18.015055 32403 solver.cpp:406] Test net output #141: loss3/loss17 = 0.0693772 (* 0.0909091 = 0.00630702 loss)
I0607 04:23:18.015072 32403 solver.cpp:406] Test net output #142: loss3/loss18 = 0.0453424 (* 0.0909091 = 0.00412204 loss)
I0607 04:23:18.015086 32403 solver.cpp:406] Test net output #143: loss3/loss19 = 0.0436533 (* 0.0909091 = 0.00396848 loss)
I0607 04:23:18.015100 32403 solver.cpp:406] Test net output #144: loss3/loss20 = 0.016143 (* 0.0909091 = 0.00146755 loss)
I0607 04:23:18.015115 32403 solver.cpp:406] Test net output #145: loss3/loss21 = 0.00571416 (* 0.0909091 = 0.000519469 loss)
I0607 04:23:18.015128 32403 solver.cpp:406] Test net output #146: loss3/loss22 = 0.00126405 (* 0.0909091 = 0.000114913 loss)
I0607 04:23:18.015141 32403 solver.cpp:406] Test net output #147: total_accuracy = 0.48
I0607 04:23:18.015151 32403 solver.cpp:406] Test net output #148: total_accuracy_not_rec = 0.468
I0607 04:23:18.015163 32403 solver.cpp:406] Test net output #149: total_confidence = 0.386788
I0607 04:23:18.015184 32403 solver.cpp:406] Test net output #150: total_confidence_not_rec = 0.33529
I0607 04:23:18.373320 32403 solver.cpp:229] Iteration 15000, loss = 3.93725
I0607 04:23:18.373420 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.433333
I0607 04:23:18.373440 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.875
I0607 04:23:18.373455 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.75
I0607 04:23:18.373476 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.625
I0607 04:23:18.373489 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.25
I0607 04:23:18.373503 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.375
I0607 04:23:18.373517 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.5
I0607 04:23:18.373529 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.75
I0607 04:23:18.373543 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.875
I0607 04:23:18.373555 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.75
I0607 04:23:18.373569 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.75
I0607 04:23:18.373582 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.875
I0607 04:23:18.373595 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 04:23:18.373608 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.875
I0607 04:23:18.373620 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.875
I0607 04:23:18.373633 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 0.875
I0607 04:23:18.373646 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 0.875
I0607 04:23:18.373657 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 0.875
I0607 04:23:18.373670 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 04:23:18.373683 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 04:23:18.373695 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 04:23:18.373708 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 04:23:18.373719 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 04:23:18.373735 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.795455
I0607 04:23:18.373749 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.7
I0607 04:23:18.373765 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.77432 (* 0.3 = 0.532295 loss)
I0607 04:23:18.373780 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.653819 (* 0.3 = 0.196146 loss)
I0607 04:23:18.373795 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.786456 (* 0.0272727 = 0.0214488 loss)
I0607 04:23:18.373811 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.07591 (* 0.0272727 = 0.029343 loss)
I0607 04:23:18.373824 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 2.3016 (* 0.0272727 = 0.062771 loss)
I0607 04:23:18.373838 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.93687 (* 0.0272727 = 0.0528238 loss)
I0607 04:23:18.373853 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 2.09605 (* 0.0272727 = 0.0571651 loss)
I0607 04:23:18.373867 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.4061 (* 0.0272727 = 0.0383482 loss)
I0607 04:23:18.373881 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 0.955398 (* 0.0272727 = 0.0260563 loss)
I0607 04:23:18.373896 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.52933 (* 0.0272727 = 0.0144363 loss)
I0607 04:23:18.373910 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.911031 (* 0.0272727 = 0.0248463 loss)
I0607 04:23:18.373924 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.707977 (* 0.0272727 = 0.0193085 loss)
I0607 04:23:18.373939 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.323274 (* 0.0272727 = 0.00881655 loss)
I0607 04:23:18.373994 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.337916 (* 0.0272727 = 0.00921588 loss)
I0607 04:23:18.374011 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.243862 (* 0.0272727 = 0.00665079 loss)
I0607 04:23:18.374025 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.295751 (* 0.0272727 = 0.00806593 loss)
I0607 04:23:18.374040 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.372785 (* 0.0272727 = 0.0101669 loss)
I0607 04:23:18.374054 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.287203 (* 0.0272727 = 0.00783281 loss)
I0607 04:23:18.374069 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.574771 (* 0.0272727 = 0.0156756 loss)
I0607 04:23:18.374084 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.03357 (* 0.0272727 = 0.000915547 loss)
I0607 04:23:18.374099 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.0165064 (* 0.0272727 = 0.000450174 loss)
I0607 04:23:18.374114 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.0028498 (* 0.0272727 = 7.77218e-05 loss)
I0607 04:23:18.374128 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.00226252 (* 0.0272727 = 6.1705e-05 loss)
I0607 04:23:18.374142 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.000751037 (* 0.0272727 = 2.04828e-05 loss)
I0607 04:23:18.374155 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.716667
I0607 04:23:18.374168 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.875
I0607 04:23:18.374181 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.75
I0607 04:23:18.374193 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.625
I0607 04:23:18.374205 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.75
I0607 04:23:18.374217 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.5
I0607 04:23:18.374229 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.625
I0607 04:23:18.374241 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.75
I0607 04:23:18.374253 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.875
I0607 04:23:18.374265 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.75
I0607 04:23:18.374279 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.75
I0607 04:23:18.374290 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.875
I0607 04:23:18.374302 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.875
I0607 04:23:18.374315 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.875
I0607 04:23:18.374326 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.875
I0607 04:23:18.374338 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 0.875
I0607 04:23:18.374351 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 0.875
I0607 04:23:18.374363 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 0.875
I0607 04:23:18.374375 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 04:23:18.374387 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 04:23:18.374399 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 04:23:18.374411 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 04:23:18.374423 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 04:23:18.374435 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.903409
I0607 04:23:18.374447 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.833333
I0607 04:23:18.374461 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.13473 (* 0.3 = 0.34042 loss)
I0607 04:23:18.374475 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.398717 (* 0.3 = 0.119615 loss)
I0607 04:23:18.374501 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.381499 (* 0.0272727 = 0.0104045 loss)
I0607 04:23:18.374513 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.588599 (* 0.0272727 = 0.0160527 loss)
I0607 04:23:18.374523 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 1.18977 (* 0.0272727 = 0.0324482 loss)
I0607 04:23:18.374539 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.03866 (* 0.0272727 = 0.028327 loss)
I0607 04:23:18.374553 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.36408 (* 0.0272727 = 0.037202 loss)
I0607 04:23:18.374567 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.13885 (* 0.0272727 = 0.0310596 loss)
I0607 04:23:18.374582 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.972432 (* 0.0272727 = 0.0265209 loss)
I0607 04:23:18.374596 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.69551 (* 0.0272727 = 0.0189685 loss)
I0607 04:23:18.374610 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 1.04199 (* 0.0272727 = 0.028418 loss)
I0607 04:23:18.374624 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.821163 (* 0.0272727 = 0.0223954 loss)
I0607 04:23:18.374639 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.215268 (* 0.0272727 = 0.00587094 loss)
I0607 04:23:18.374652 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.468924 (* 0.0272727 = 0.0127888 loss)
I0607 04:23:18.374667 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.251367 (* 0.0272727 = 0.00685546 loss)
I0607 04:23:18.374681 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.333219 (* 0.0272727 = 0.0090878 loss)
I0607 04:23:18.374696 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.324527 (* 0.0272727 = 0.00885075 loss)
I0607 04:23:18.374711 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.333339 (* 0.0272727 = 0.00909107 loss)
I0607 04:23:18.374724 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.478847 (* 0.0272727 = 0.0130595 loss)
I0607 04:23:18.374739 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.0379018 (* 0.0272727 = 0.00103369 loss)
I0607 04:23:18.374754 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.026734 (* 0.0272727 = 0.00072911 loss)
I0607 04:23:18.374768 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.0181035 (* 0.0272727 = 0.000493733 loss)
I0607 04:23:18.374785 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.00326278 (* 0.0272727 = 8.89848e-05 loss)
I0607 04:23:18.374801 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 0.00154007 (* 0.0272727 = 4.2002e-05 loss)
I0607 04:23:18.374814 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.866667
I0607 04:23:18.374827 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 0.875
I0607 04:23:18.374840 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 1
I0607 04:23:18.374851 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 1
I0607 04:23:18.374862 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 1
I0607 04:23:18.374874 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.875
I0607 04:23:18.374886 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 1
I0607 04:23:18.374898 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.875
I0607 04:23:18.374910 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.875
I0607 04:23:18.374923 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.75
I0607 04:23:18.374935 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.75
I0607 04:23:18.374948 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 1
I0607 04:23:18.374959 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.875
I0607 04:23:18.374971 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 0.875
I0607 04:23:18.374994 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 0.875
I0607 04:23:18.375007 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 0.875
I0607 04:23:18.375020 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 0.875
I0607 04:23:18.375031 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 0.875
I0607 04:23:18.375043 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 04:23:18.375056 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 04:23:18.375067 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 04:23:18.375079 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 04:23:18.375092 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 04:23:18.375103 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.943182
I0607 04:23:18.375115 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.966667
I0607 04:23:18.375130 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.555058 (* 1 = 0.555058 loss)
I0607 04:23:18.375145 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.206603 (* 1 = 0.206603 loss)
I0607 04:23:18.375159 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.273126 (* 0.0909091 = 0.0248297 loss)
I0607 04:23:18.375174 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.0672972 (* 0.0909091 = 0.00611793 loss)
I0607 04:23:18.375188 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.113533 (* 0.0909091 = 0.0103212 loss)
I0607 04:23:18.375202 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.249048 (* 0.0909091 = 0.0226408 loss)
I0607 04:23:18.375216 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.495266 (* 0.0909091 = 0.0450242 loss)
I0607 04:23:18.375231 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.266551 (* 0.0909091 = 0.0242319 loss)
I0607 04:23:18.375246 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.416639 (* 0.0909091 = 0.0378762 loss)
I0607 04:23:18.375260 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.17876 (* 0.0909091 = 0.0162509 loss)
I0607 04:23:18.375270 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.650025 (* 0.0909091 = 0.0590932 loss)
I0607 04:23:18.375284 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.621367 (* 0.0909091 = 0.0564879 loss)
I0607 04:23:18.375299 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.128037 (* 0.0909091 = 0.0116397 loss)
I0607 04:23:18.375313 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.49141 (* 0.0909091 = 0.0446737 loss)
I0607 04:23:18.375327 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.150416 (* 0.0909091 = 0.0136742 loss)
I0607 04:23:18.375342 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.269793 (* 0.0909091 = 0.0245266 loss)
I0607 04:23:18.375356 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.301367 (* 0.0909091 = 0.027397 loss)
I0607 04:23:18.375370 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.33988 (* 0.0909091 = 0.0308981 loss)
I0607 04:23:18.375385 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.507737 (* 0.0909091 = 0.0461579 loss)
I0607 04:23:18.375399 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.00830534 (* 0.0909091 = 0.000755031 loss)
I0607 04:23:18.375413 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.00101596 (* 0.0909091 = 9.23603e-05 loss)
I0607 04:23:18.375427 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.000636749 (* 0.0909091 = 5.78863e-05 loss)
I0607 04:23:18.375442 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.000323224 (* 0.0909091 = 2.9384e-05 loss)
I0607 04:23:18.375457 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 8.02988e-05 (* 0.0909091 = 7.29989e-06 loss)
I0607 04:23:18.375479 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.625
I0607 04:23:18.375493 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.75
I0607 04:23:18.375504 32403 solver.cpp:245] Train net output #149: total_confidence = 0.547797
I0607 04:23:18.375516 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.473845
I0607 04:23:18.375530 32403 sgd_solver.cpp:106] Iteration 15000, lr = 0.001
I0607 04:24:58.407971 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.5663 > 30) by scale factor 0.981474
I0607 04:25:06.906633 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 50.6246 > 30) by scale factor 0.592598
I0607 04:25:48.678412 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.6243 > 30) by scale factor 0.866443
I0607 04:26:10.308965 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.7732 > 30) by scale factor 0.888279
I0607 04:27:21.393504 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.8349 > 30) by scale factor 0.792919
I0607 04:27:32.215299 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 35.884 > 30) by scale factor 0.836028
I0607 04:29:28.046351 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.3523 > 30) by scale factor 0.956867
I0607 04:29:44.682868 32403 solver.cpp:229] Iteration 15500, loss = 3.96013
I0607 04:29:44.682936 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.588235
I0607 04:29:44.682955 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.625
I0607 04:29:44.682970 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.75
I0607 04:29:44.682982 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.75
I0607 04:29:44.682996 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.5
I0607 04:29:44.683008 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.125
I0607 04:29:44.683022 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.75
I0607 04:29:44.683034 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.875
I0607 04:29:44.683048 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.875
I0607 04:29:44.683059 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.875
I0607 04:29:44.683073 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 1
I0607 04:29:44.683086 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 1
I0607 04:29:44.683099 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 1
I0607 04:29:44.683111 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 1
I0607 04:29:44.683123 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 1
I0607 04:29:44.683135 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 04:29:44.683147 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 04:29:44.683159 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 04:29:44.683171 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 04:29:44.683183 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 04:29:44.683195 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 04:29:44.683207 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 04:29:44.683219 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 04:29:44.683233 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.875
I0607 04:29:44.683244 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.823529
I0607 04:29:44.683261 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.45329 (* 0.3 = 0.435988 loss)
I0607 04:29:44.683276 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.461958 (* 0.3 = 0.138587 loss)
I0607 04:29:44.683291 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.920772 (* 0.0272727 = 0.025112 loss)
I0607 04:29:44.683305 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 0.674306 (* 0.0272727 = 0.0183902 loss)
I0607 04:29:44.683320 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.58919 (* 0.0272727 = 0.0433416 loss)
I0607 04:29:44.683334 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.46205 (* 0.0272727 = 0.039874 loss)
I0607 04:29:44.683348 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 2.79611 (* 0.0272727 = 0.0762575 loss)
I0607 04:29:44.683362 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 0.893843 (* 0.0272727 = 0.0243775 loss)
I0607 04:29:44.683377 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 0.76652 (* 0.0272727 = 0.0209051 loss)
I0607 04:29:44.683390 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.479679 (* 0.0272727 = 0.0130821 loss)
I0607 04:29:44.683405 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.514473 (* 0.0272727 = 0.0140311 loss)
I0607 04:29:44.683420 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.0668805 (* 0.0272727 = 0.00182401 loss)
I0607 04:29:44.683434 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.0060082 (* 0.0272727 = 0.00016386 loss)
I0607 04:29:44.683450 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.00297562 (* 0.0272727 = 8.11533e-05 loss)
I0607 04:29:44.683506 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.0022162 (* 0.0272727 = 6.04418e-05 loss)
I0607 04:29:44.683521 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.00149148 (* 0.0272727 = 4.06768e-05 loss)
I0607 04:29:44.683537 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.000610505 (* 0.0272727 = 1.66501e-05 loss)
I0607 04:29:44.683552 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.000444484 (* 0.0272727 = 1.21223e-05 loss)
I0607 04:29:44.683565 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.00014851 (* 0.0272727 = 4.05026e-06 loss)
I0607 04:29:44.683579 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 7.39157e-05 (* 0.0272727 = 2.01588e-06 loss)
I0607 04:29:44.683594 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 2.21375e-05 (* 0.0272727 = 6.03751e-07 loss)
I0607 04:29:44.683612 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 2.56698e-05 (* 0.0272727 = 7.00085e-07 loss)
I0607 04:29:44.683627 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 1.51405e-05 (* 0.0272727 = 4.12922e-07 loss)
I0607 04:29:44.683641 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 1.79571e-05 (* 0.0272727 = 4.8974e-07 loss)
I0607 04:29:44.683655 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.705882
I0607 04:29:44.683666 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 1
I0607 04:29:44.683678 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.75
I0607 04:29:44.683691 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.875
I0607 04:29:44.683703 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.75
I0607 04:29:44.683715 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.625
I0607 04:29:44.683727 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.625
I0607 04:29:44.683739 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.875
I0607 04:29:44.683751 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.875
I0607 04:29:44.683766 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.875
I0607 04:29:44.683779 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 1
I0607 04:29:44.683790 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 1
I0607 04:29:44.683802 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 1
I0607 04:29:44.683815 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 1
I0607 04:29:44.683826 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 1
I0607 04:29:44.683837 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 04:29:44.683850 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 04:29:44.683861 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 04:29:44.683873 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 04:29:44.683886 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 04:29:44.683897 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 04:29:44.683908 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 04:29:44.683920 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 04:29:44.683933 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.914773
I0607 04:29:44.683944 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.941176
I0607 04:29:44.683959 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 0.924244 (* 0.3 = 0.277273 loss)
I0607 04:29:44.683974 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.290265 (* 0.3 = 0.0870795 loss)
I0607 04:29:44.683989 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.212912 (* 0.0272727 = 0.00580668 loss)
I0607 04:29:44.684015 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.690123 (* 0.0272727 = 0.0188215 loss)
I0607 04:29:44.684029 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.524549 (* 0.0272727 = 0.0143059 loss)
I0607 04:29:44.684043 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.47925 (* 0.0272727 = 0.0403433 loss)
I0607 04:29:44.684057 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.84631 (* 0.0272727 = 0.050354 loss)
I0607 04:29:44.684072 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.19907 (* 0.0272727 = 0.032702 loss)
I0607 04:29:44.684085 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.494041 (* 0.0272727 = 0.0134738 loss)
I0607 04:29:44.684100 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.618256 (* 0.0272727 = 0.0168615 loss)
I0607 04:29:44.684114 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.343376 (* 0.0272727 = 0.00936479 loss)
I0607 04:29:44.684128 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.0403036 (* 0.0272727 = 0.00109919 loss)
I0607 04:29:44.684142 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.00746755 (* 0.0272727 = 0.00020366 loss)
I0607 04:29:44.684157 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.00288963 (* 0.0272727 = 7.8808e-05 loss)
I0607 04:29:44.684171 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.00318157 (* 0.0272727 = 8.67702e-05 loss)
I0607 04:29:44.684185 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.00189538 (* 0.0272727 = 5.16921e-05 loss)
I0607 04:29:44.684200 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.00160461 (* 0.0272727 = 4.37622e-05 loss)
I0607 04:29:44.684214 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.00110714 (* 0.0272727 = 3.01946e-05 loss)
I0607 04:29:44.684228 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.000576916 (* 0.0272727 = 1.57341e-05 loss)
I0607 04:29:44.684242 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.000716813 (* 0.0272727 = 1.95494e-05 loss)
I0607 04:29:44.684257 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.000168361 (* 0.0272727 = 4.59165e-06 loss)
I0607 04:29:44.684270 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.000130795 (* 0.0272727 = 3.56713e-06 loss)
I0607 04:29:44.684284 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.000112799 (* 0.0272727 = 3.07633e-06 loss)
I0607 04:29:44.684299 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 0.000157395 (* 0.0272727 = 4.2926e-06 loss)
I0607 04:29:44.684311 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.901961
I0607 04:29:44.684324 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 04:29:44.684335 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 1
I0607 04:29:44.684347 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.875
I0607 04:29:44.684360 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.875
I0607 04:29:44.684372 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.75
I0607 04:29:44.684384 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.75
I0607 04:29:44.684396 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.875
I0607 04:29:44.684408 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 1
I0607 04:29:44.684420 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 1
I0607 04:29:44.684432 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 1
I0607 04:29:44.684443 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 1
I0607 04:29:44.684455 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 1
I0607 04:29:44.684466 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 1
I0607 04:29:44.684479 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 04:29:44.684490 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 04:29:44.684512 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 04:29:44.684525 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 04:29:44.684538 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 04:29:44.684550 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 04:29:44.684561 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 04:29:44.684572 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 04:29:44.684584 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 04:29:44.684597 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.965909
I0607 04:29:44.684608 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.980392
I0607 04:29:44.684623 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.541388 (* 1 = 0.541388 loss)
I0607 04:29:44.684636 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.170911 (* 1 = 0.170911 loss)
I0607 04:29:44.684654 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.048757 (* 0.0909091 = 0.00443246 loss)
I0607 04:29:44.684670 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.0776325 (* 0.0909091 = 0.0070575 loss)
I0607 04:29:44.684687 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.281627 (* 0.0909091 = 0.0256025 loss)
I0607 04:29:44.684697 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.711952 (* 0.0909091 = 0.0647229 loss)
I0607 04:29:44.684725 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 1.58975 (* 0.0909091 = 0.144523 loss)
I0607 04:29:44.684751 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.692893 (* 0.0909091 = 0.0629903 loss)
I0607 04:29:44.684767 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.30739 (* 0.0909091 = 0.0279446 loss)
I0607 04:29:44.684782 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.246524 (* 0.0909091 = 0.0224112 loss)
I0607 04:29:44.684797 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.0891338 (* 0.0909091 = 0.00810307 loss)
I0607 04:29:44.684813 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.00374389 (* 0.0909091 = 0.000340353 loss)
I0607 04:29:44.684828 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.000294385 (* 0.0909091 = 2.67622e-05 loss)
I0607 04:29:44.684842 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 8.77461e-05 (* 0.0909091 = 7.97692e-06 loss)
I0607 04:29:44.684856 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 3.95287e-05 (* 0.0909091 = 3.59352e-06 loss)
I0607 04:29:44.684870 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 3.09749e-05 (* 0.0909091 = 2.8159e-06 loss)
I0607 04:29:44.684885 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 7.2208e-05 (* 0.0909091 = 6.56436e-06 loss)
I0607 04:29:44.684900 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 5.61219e-05 (* 0.0909091 = 5.10199e-06 loss)
I0607 04:29:44.684913 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 8.62793e-06 (* 0.0909091 = 7.84357e-07 loss)
I0607 04:29:44.684927 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 3.69552e-06 (* 0.0909091 = 3.35956e-07 loss)
I0607 04:29:44.684942 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 3.05475e-06 (* 0.0909091 = 2.77705e-07 loss)
I0607 04:29:44.684957 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 1.34111e-06 (* 0.0909091 = 1.21919e-07 loss)
I0607 04:29:44.684970 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 1.50502e-06 (* 0.0909091 = 1.3682e-07 loss)
I0607 04:29:44.684984 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 2.23518e-06 (* 0.0909091 = 2.03198e-07 loss)
I0607 04:29:44.684996 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.625
I0607 04:29:44.685020 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.5
I0607 04:29:44.685034 32403 solver.cpp:245] Train net output #149: total_confidence = 0.538222
I0607 04:29:44.685045 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.503105
I0607 04:29:44.685058 32403 sgd_solver.cpp:106] Iteration 15500, lr = 0.001
I0607 04:29:57.407649 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 35.1197 > 30) by scale factor 0.854221
I0607 04:29:59.731894 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.7346 > 30) by scale factor 0.795026
I0607 04:30:21.364296 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.3876 > 30) by scale factor 0.987245
I0607 04:32:28.775212 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.3261 > 30) by scale factor 0.989246
I0607 04:32:36.491792 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 35.6019 > 30) by scale factor 0.842651
I0607 04:33:06.585661 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 39.1298 > 30) by scale factor 0.766679
I0607 04:33:29.769191 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.106 > 30) by scale factor 0.906181
I0607 04:33:31.311965 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.2311 > 30) by scale factor 0.805779
I0607 04:34:50.124828 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.0006 > 30) by scale factor 0.967722
I0607 04:35:32.607641 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 36.0022 > 30) by scale factor 0.833283
I0607 04:35:38.016664 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.7177 > 30) by scale factor 0.88974
I0607 04:36:10.875344 32403 solver.cpp:229] Iteration 16000, loss = 4.06033
I0607 04:36:10.875432 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.426667
I0607 04:36:10.875452 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 1
I0607 04:36:10.875465 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.75
I0607 04:36:10.875478 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.375
I0607 04:36:10.875491 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.25
I0607 04:36:10.875504 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.375
I0607 04:36:10.875521 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.5
I0607 04:36:10.875535 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.5
I0607 04:36:10.875547 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.625
I0607 04:36:10.875560 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.875
I0607 04:36:10.875574 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.75
I0607 04:36:10.875587 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.625
I0607 04:36:10.875600 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.75
I0607 04:36:10.875613 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.875
I0607 04:36:10.875625 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.75
I0607 04:36:10.875638 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 0.75
I0607 04:36:10.875650 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 0.75
I0607 04:36:10.875663 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 04:36:10.875674 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 04:36:10.875686 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 04:36:10.875699 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 04:36:10.875710 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 04:36:10.875722 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 04:36:10.875735 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.732955
I0607 04:36:10.875746 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.733333
I0607 04:36:10.875766 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.60536 (* 0.3 = 0.481609 loss)
I0607 04:36:10.875782 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.747253 (* 0.3 = 0.224176 loss)
I0607 04:36:10.875797 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.211161 (* 0.0272727 = 0.00575892 loss)
I0607 04:36:10.875813 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 0.702959 (* 0.0272727 = 0.0191716 loss)
I0607 04:36:10.875826 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.76735 (* 0.0272727 = 0.0482005 loss)
I0607 04:36:10.875840 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.94168 (* 0.0272727 = 0.0529549 loss)
I0607 04:36:10.875854 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.9086 (* 0.0272727 = 0.0520527 loss)
I0607 04:36:10.875869 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.48295 (* 0.0272727 = 0.0404441 loss)
I0607 04:36:10.875882 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 1.4962 (* 0.0272727 = 0.0408053 loss)
I0607 04:36:10.875896 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 1.36383 (* 0.0272727 = 0.0371953 loss)
I0607 04:36:10.875910 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.553547 (* 0.0272727 = 0.0150967 loss)
I0607 04:36:10.875924 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.887395 (* 0.0272727 = 0.0242017 loss)
I0607 04:36:10.875938 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 1.18504 (* 0.0272727 = 0.0323191 loss)
I0607 04:36:10.875952 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.716411 (* 0.0272727 = 0.0195385 loss)
I0607 04:36:10.875986 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.570678 (* 0.0272727 = 0.0155639 loss)
I0607 04:36:10.876003 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.54964 (* 0.0272727 = 0.0149902 loss)
I0607 04:36:10.876016 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.556371 (* 0.0272727 = 0.0151738 loss)
I0607 04:36:10.876030 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.819685 (* 0.0272727 = 0.022355 loss)
I0607 04:36:10.876044 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.0116467 (* 0.0272727 = 0.000317636 loss)
I0607 04:36:10.876060 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.00191614 (* 0.0272727 = 5.22584e-05 loss)
I0607 04:36:10.876073 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.000342949 (* 0.0272727 = 9.35316e-06 loss)
I0607 04:36:10.876088 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.000246066 (* 0.0272727 = 6.7109e-06 loss)
I0607 04:36:10.876102 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 7.9363e-05 (* 0.0272727 = 2.16444e-06 loss)
I0607 04:36:10.876116 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 5.86882e-05 (* 0.0272727 = 1.60059e-06 loss)
I0607 04:36:10.876128 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.52
I0607 04:36:10.876142 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 1
I0607 04:36:10.876154 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.875
I0607 04:36:10.876166 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.5
I0607 04:36:10.876178 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.625
I0607 04:36:10.876190 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.25
I0607 04:36:10.876202 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.625
I0607 04:36:10.876214 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.375
I0607 04:36:10.876227 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.625
I0607 04:36:10.876240 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.75
I0607 04:36:10.876251 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.625
I0607 04:36:10.876263 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.75
I0607 04:36:10.876276 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.75
I0607 04:36:10.876287 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.875
I0607 04:36:10.876299 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.75
I0607 04:36:10.876312 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 0.875
I0607 04:36:10.876323 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 0.75
I0607 04:36:10.876337 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 04:36:10.876348 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 04:36:10.876359 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 04:36:10.876371 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 04:36:10.876384 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 04:36:10.876395 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 04:36:10.876407 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.789773
I0607 04:36:10.876420 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.813333
I0607 04:36:10.876433 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.28426 (* 0.3 = 0.385277 loss)
I0607 04:36:10.876447 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.584306 (* 0.3 = 0.175292 loss)
I0607 04:36:10.876461 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.0990986 (* 0.0272727 = 0.00270269 loss)
I0607 04:36:10.876477 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.386798 (* 0.0272727 = 0.010549 loss)
I0607 04:36:10.876502 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 1.14327 (* 0.0272727 = 0.0311801 loss)
I0607 04:36:10.876523 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.21571 (* 0.0272727 = 0.0331558 loss)
I0607 04:36:10.876548 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 2.49653 (* 0.0272727 = 0.0680873 loss)
I0607 04:36:10.876566 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.25021 (* 0.0272727 = 0.0340965 loss)
I0607 04:36:10.876581 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 1.14954 (* 0.0272727 = 0.0313511 loss)
I0607 04:36:10.876596 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 1.06778 (* 0.0272727 = 0.0291212 loss)
I0607 04:36:10.876610 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.86438 (* 0.0272727 = 0.023574 loss)
I0607 04:36:10.876624 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 1.0654 (* 0.0272727 = 0.0290562 loss)
I0607 04:36:10.876637 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 1.1463 (* 0.0272727 = 0.0312626 loss)
I0607 04:36:10.876652 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.617177 (* 0.0272727 = 0.0168321 loss)
I0607 04:36:10.876665 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.508327 (* 0.0272727 = 0.0138635 loss)
I0607 04:36:10.876680 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.65537 (* 0.0272727 = 0.0178737 loss)
I0607 04:36:10.876690 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.57367 (* 0.0272727 = 0.0156455 loss)
I0607 04:36:10.876705 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 1.16564 (* 0.0272727 = 0.0317901 loss)
I0607 04:36:10.876719 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.00382467 (* 0.0272727 = 0.000104309 loss)
I0607 04:36:10.876734 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.000744087 (* 0.0272727 = 2.02933e-05 loss)
I0607 04:36:10.876749 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.000164977 (* 0.0272727 = 4.49937e-06 loss)
I0607 04:36:10.876762 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 6.38556e-05 (* 0.0272727 = 1.74152e-06 loss)
I0607 04:36:10.876777 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.000104881 (* 0.0272727 = 2.8604e-06 loss)
I0607 04:36:10.876791 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 0.00012856 (* 0.0272727 = 3.50617e-06 loss)
I0607 04:36:10.876803 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.733333
I0607 04:36:10.876818 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 04:36:10.876830 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.875
I0607 04:36:10.876842 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.75
I0607 04:36:10.876855 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.75
I0607 04:36:10.876868 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.625
I0607 04:36:10.876879 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.75
I0607 04:36:10.876893 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.75
I0607 04:36:10.876904 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.875
I0607 04:36:10.876916 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.875
I0607 04:36:10.876929 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.875
I0607 04:36:10.876940 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.875
I0607 04:36:10.876952 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.75
I0607 04:36:10.876965 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 0.875
I0607 04:36:10.876976 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 0.75
I0607 04:36:10.876987 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 0.875
I0607 04:36:10.877012 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 0.875
I0607 04:36:10.877025 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 04:36:10.877038 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 04:36:10.877049 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 04:36:10.877061 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 04:36:10.877074 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 04:36:10.877084 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 04:36:10.877096 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.880682
I0607 04:36:10.877109 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.893333
I0607 04:36:10.877140 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.876464 (* 1 = 0.876464 loss)
I0607 04:36:10.877157 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.399421 (* 1 = 0.399421 loss)
I0607 04:36:10.877172 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.0504953 (* 0.0909091 = 0.00459048 loss)
I0607 04:36:10.877187 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.394226 (* 0.0909091 = 0.0358388 loss)
I0607 04:36:10.877202 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.950872 (* 0.0909091 = 0.0864429 loss)
I0607 04:36:10.877215 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.840954 (* 0.0909091 = 0.0764504 loss)
I0607 04:36:10.877229 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 1.15247 (* 0.0909091 = 0.10477 loss)
I0607 04:36:10.877243 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.82677 (* 0.0909091 = 0.075161 loss)
I0607 04:36:10.877257 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.916955 (* 0.0909091 = 0.0833596 loss)
I0607 04:36:10.877271 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.481739 (* 0.0909091 = 0.0437944 loss)
I0607 04:36:10.877285 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.52712 (* 0.0909091 = 0.04792 loss)
I0607 04:36:10.877300 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.460913 (* 0.0909091 = 0.0419012 loss)
I0607 04:36:10.877313 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.493546 (* 0.0909091 = 0.0448678 loss)
I0607 04:36:10.877327 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.602142 (* 0.0909091 = 0.0547402 loss)
I0607 04:36:10.877341 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.380436 (* 0.0909091 = 0.0345851 loss)
I0607 04:36:10.877356 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.396685 (* 0.0909091 = 0.0360622 loss)
I0607 04:36:10.877369 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.460038 (* 0.0909091 = 0.0418217 loss)
I0607 04:36:10.877383 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.414051 (* 0.0909091 = 0.037641 loss)
I0607 04:36:10.877393 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.0101608 (* 0.0909091 = 0.000923712 loss)
I0607 04:36:10.877409 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.00111466 (* 0.0909091 = 0.000101332 loss)
I0607 04:36:10.877424 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.000450791 (* 0.0909091 = 4.0981e-05 loss)
I0607 04:36:10.877439 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.000155955 (* 0.0909091 = 1.41777e-05 loss)
I0607 04:36:10.877452 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.000100264 (* 0.0909091 = 9.11489e-06 loss)
I0607 04:36:10.877466 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 2.81044e-05 (* 0.0909091 = 2.55494e-06 loss)
I0607 04:36:10.877478 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.25
I0607 04:36:10.877490 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.125
I0607 04:36:10.877514 32403 solver.cpp:245] Train net output #149: total_confidence = 0.201909
I0607 04:36:10.877527 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.192204
I0607 04:36:10.877542 32403 sgd_solver.cpp:106] Iteration 16000, lr = 0.001
I0607 04:36:27.465744 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.7654 > 30) by scale factor 0.862926
I0607 04:36:54.468895 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.619 > 30) by scale factor 0.979785
I0607 04:36:59.870232 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.7033 > 30) by scale factor 0.864471
I0607 04:37:02.192540 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.1741 > 30) by scale factor 0.932428
I0607 04:37:39.257208 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.0483 > 30) by scale factor 0.998393
I0607 04:38:41.833240 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 39.6705 > 30) by scale factor 0.756229
I0607 04:40:27.566345 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.5579 > 30) by scale factor 0.798767
I0607 04:40:32.998250 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 39.9112 > 30) by scale factor 0.751668
I0607 04:41:27.042598 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.1291 > 30) by scale factor 0.995714
I0607 04:42:20.357388 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.5274 > 30) by scale factor 0.894792
I0607 04:42:36.998040 32403 solver.cpp:229] Iteration 16500, loss = 3.84747
I0607 04:42:36.998116 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.4375
I0607 04:42:36.998134 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.625
I0607 04:42:36.998148 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0
I0607 04:42:36.998169 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.5
I0607 04:42:36.998183 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.75
I0607 04:42:36.998196 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.625
I0607 04:42:36.998209 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.5
I0607 04:42:36.998222 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.625
I0607 04:42:36.998235 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.875
I0607 04:42:36.998251 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.875
I0607 04:42:36.998265 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.875
I0607 04:42:36.998278 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 1
I0607 04:42:36.998291 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 1
I0607 04:42:36.998304 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 1
I0607 04:42:36.998317 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 1
I0607 04:42:36.998329 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 04:42:36.998340 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 04:42:36.998353 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 04:42:36.998364 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 04:42:36.998376 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 04:42:36.998389 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 04:42:36.998400 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 04:42:36.998412 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 04:42:36.998425 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.829545
I0607 04:42:36.998445 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.729167
I0607 04:42:36.998463 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.83398 (* 0.3 = 0.550193 loss)
I0607 04:42:36.998478 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.604494 (* 0.3 = 0.181348 loss)
I0607 04:42:36.998493 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 1.51596 (* 0.0272727 = 0.0413445 loss)
I0607 04:42:36.998507 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 2.43811 (* 0.0272727 = 0.0664939 loss)
I0607 04:42:36.998522 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.7503 (* 0.0272727 = 0.0477355 loss)
I0607 04:42:36.998535 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.20689 (* 0.0272727 = 0.0329153 loss)
I0607 04:42:36.998549 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.67587 (* 0.0272727 = 0.0457055 loss)
I0607 04:42:36.998564 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.52458 (* 0.0272727 = 0.0415796 loss)
I0607 04:42:36.998577 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 0.949103 (* 0.0272727 = 0.0258846 loss)
I0607 04:42:36.998599 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.849621 (* 0.0272727 = 0.0231715 loss)
I0607 04:42:36.998613 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.779857 (* 0.0272727 = 0.0212688 loss)
I0607 04:42:36.998627 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.699788 (* 0.0272727 = 0.0190851 loss)
I0607 04:42:36.998642 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.0706271 (* 0.0272727 = 0.00192619 loss)
I0607 04:42:36.998656 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.0393759 (* 0.0272727 = 0.00107389 loss)
I0607 04:42:36.998713 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.0342205 (* 0.0272727 = 0.000933286 loss)
I0607 04:42:36.998729 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.037603 (* 0.0272727 = 0.00102554 loss)
I0607 04:42:36.998746 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.0552034 (* 0.0272727 = 0.00150555 loss)
I0607 04:42:36.998761 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.0406769 (* 0.0272727 = 0.00110937 loss)
I0607 04:42:36.998776 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.0710497 (* 0.0272727 = 0.00193772 loss)
I0607 04:42:36.998790 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.0677442 (* 0.0272727 = 0.00184757 loss)
I0607 04:42:36.998805 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.0451244 (* 0.0272727 = 0.00123067 loss)
I0607 04:42:36.998819 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.0376981 (* 0.0272727 = 0.00102813 loss)
I0607 04:42:36.998834 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.0702412 (* 0.0272727 = 0.00191567 loss)
I0607 04:42:36.998848 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.0914238 (* 0.0272727 = 0.00249338 loss)
I0607 04:42:36.998862 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.583333
I0607 04:42:36.998874 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.625
I0607 04:42:36.998886 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.625
I0607 04:42:36.998899 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.75
I0607 04:42:36.998911 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.5
I0607 04:42:36.998931 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.5
I0607 04:42:36.998944 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.625
I0607 04:42:36.998955 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.75
I0607 04:42:36.998967 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.875
I0607 04:42:36.998988 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.875
I0607 04:42:36.999011 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.875
I0607 04:42:36.999027 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 1
I0607 04:42:36.999038 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 1
I0607 04:42:36.999050 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 1
I0607 04:42:36.999063 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 1
I0607 04:42:36.999083 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 04:42:36.999095 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 04:42:36.999107 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 04:42:36.999119 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 04:42:36.999130 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 04:42:36.999142 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 04:42:36.999155 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 04:42:36.999166 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 04:42:36.999178 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.869318
I0607 04:42:36.999191 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.708333
I0607 04:42:36.999207 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.63255 (* 0.3 = 0.489766 loss)
I0607 04:42:36.999220 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.548 (* 0.3 = 0.1644 loss)
I0607 04:42:36.999234 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 1.3242 (* 0.0272727 = 0.0361146 loss)
I0607 04:42:36.999249 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 1.63642 (* 0.0272727 = 0.0446297 loss)
I0607 04:42:36.999275 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 1.3697 (* 0.0272727 = 0.0373555 loss)
I0607 04:42:36.999291 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.53712 (* 0.0272727 = 0.0419215 loss)
I0607 04:42:36.999310 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.19849 (* 0.0272727 = 0.0326861 loss)
I0607 04:42:36.999325 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.45399 (* 0.0272727 = 0.0396542 loss)
I0607 04:42:36.999342 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.963545 (* 0.0272727 = 0.0262785 loss)
I0607 04:42:36.999357 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.716877 (* 0.0272727 = 0.0195512 loss)
I0607 04:42:36.999372 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.654524 (* 0.0272727 = 0.0178507 loss)
I0607 04:42:36.999382 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.69035 (* 0.0272727 = 0.0188277 loss)
I0607 04:42:36.999392 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.104953 (* 0.0272727 = 0.00286234 loss)
I0607 04:42:36.999413 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.0557957 (* 0.0272727 = 0.0015217 loss)
I0607 04:42:36.999428 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.0427762 (* 0.0272727 = 0.00116662 loss)
I0607 04:42:36.999442 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.0459444 (* 0.0272727 = 0.00125303 loss)
I0607 04:42:36.999456 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.0357803 (* 0.0272727 = 0.000975826 loss)
I0607 04:42:36.999471 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.0337601 (* 0.0272727 = 0.000920729 loss)
I0607 04:42:36.999485 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.0309436 (* 0.0272727 = 0.000843916 loss)
I0607 04:42:36.999500 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.0497658 (* 0.0272727 = 0.00135725 loss)
I0607 04:42:36.999514 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.0438335 (* 0.0272727 = 0.00119546 loss)
I0607 04:42:36.999528 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.0563136 (* 0.0272727 = 0.00153582 loss)
I0607 04:42:36.999543 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.0579293 (* 0.0272727 = 0.00157989 loss)
I0607 04:42:36.999557 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 0.0598274 (* 0.0272727 = 0.00163166 loss)
I0607 04:42:36.999569 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.625
I0607 04:42:36.999583 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 0.625
I0607 04:42:36.999594 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.5
I0607 04:42:36.999606 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.75
I0607 04:42:36.999619 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.875
I0607 04:42:36.999631 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.75
I0607 04:42:36.999644 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.875
I0607 04:42:36.999655 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.625
I0607 04:42:36.999667 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.75
I0607 04:42:36.999680 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.75
I0607 04:42:36.999692 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.875
I0607 04:42:36.999704 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 1
I0607 04:42:36.999716 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 1
I0607 04:42:36.999727 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 1
I0607 04:42:36.999739 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 04:42:36.999752 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 04:42:36.999774 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 04:42:36.999788 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 04:42:36.999802 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 04:42:36.999814 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 04:42:36.999827 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 04:42:36.999840 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 04:42:36.999851 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 04:42:36.999863 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.892045
I0607 04:42:36.999876 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.75
I0607 04:42:36.999889 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 1.26594 (* 1 = 1.26594 loss)
I0607 04:42:36.999903 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.446933 (* 1 = 0.446933 loss)
I0607 04:42:36.999918 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 1.36586 (* 0.0909091 = 0.12417 loss)
I0607 04:42:36.999933 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 1.50021 (* 0.0909091 = 0.136383 loss)
I0607 04:42:36.999951 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 1.26803 (* 0.0909091 = 0.115275 loss)
I0607 04:42:36.999965 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.68155 (* 0.0909091 = 0.0619591 loss)
I0607 04:42:36.999979 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.839759 (* 0.0909091 = 0.0763418 loss)
I0607 04:42:36.999994 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.651502 (* 0.0909091 = 0.0592274 loss)
I0607 04:42:37.000015 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.96042 (* 0.0909091 = 0.087311 loss)
I0607 04:42:37.000028 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.828088 (* 0.0909091 = 0.0752808 loss)
I0607 04:42:37.000042 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.772153 (* 0.0909091 = 0.0701957 loss)
I0607 04:42:37.000057 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.785712 (* 0.0909091 = 0.0714284 loss)
I0607 04:42:37.000072 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.0467852 (* 0.0909091 = 0.0042532 loss)
I0607 04:42:37.000085 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.0335554 (* 0.0909091 = 0.00305049 loss)
I0607 04:42:37.000100 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.0196504 (* 0.0909091 = 0.0017864 loss)
I0607 04:42:37.000115 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.0185535 (* 0.0909091 = 0.00168669 loss)
I0607 04:42:37.000129 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.0119896 (* 0.0909091 = 0.00108996 loss)
I0607 04:42:37.000140 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.017407 (* 0.0909091 = 0.00158245 loss)
I0607 04:42:37.000154 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.0155733 (* 0.0909091 = 0.00141576 loss)
I0607 04:42:37.000169 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.0198148 (* 0.0909091 = 0.00180135 loss)
I0607 04:42:37.000183 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.012942 (* 0.0909091 = 0.00117654 loss)
I0607 04:42:37.000198 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.0208289 (* 0.0909091 = 0.00189354 loss)
I0607 04:42:37.000212 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.0171786 (* 0.0909091 = 0.00156169 loss)
I0607 04:42:37.000226 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 0.0132205 (* 0.0909091 = 0.00120186 loss)
I0607 04:42:37.000239 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.375
I0607 04:42:37.000252 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.25
I0607 04:42:37.000273 32403 solver.cpp:245] Train net output #149: total_confidence = 0.316382
I0607 04:42:37.000288 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.266396
I0607 04:42:37.000301 32403 sgd_solver.cpp:106] Iteration 16500, lr = 0.001
I0607 04:42:42.768522 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.598 > 30) by scale factor 0.920302
I0607 04:43:48.410701 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.4239 > 30) by scale factor 0.925244
I0607 04:45:32.657495 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 50.6676 > 30) by scale factor 0.592095
I0607 04:47:19.328470 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.0633 > 30) by scale factor 0.997895
I0607 04:49:03.163399 32403 solver.cpp:229] Iteration 17000, loss = 3.90182
I0607 04:49:03.163576 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.508772
I0607 04:49:03.163599 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.625
I0607 04:49:03.163614 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.625
I0607 04:49:03.163635 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.375
I0607 04:49:03.163648 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.5
I0607 04:49:03.163661 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.25
I0607 04:49:03.163674 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.625
I0607 04:49:03.163687 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.5
I0607 04:49:03.163702 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.625
I0607 04:49:03.163714 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.75
I0607 04:49:03.163727 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.75
I0607 04:49:03.163741 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 1
I0607 04:49:03.163754 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 04:49:03.163767 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.875
I0607 04:49:03.163779 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 1
I0607 04:49:03.163791 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 04:49:03.163803 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 04:49:03.163816 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 04:49:03.163828 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 04:49:03.163841 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 04:49:03.163852 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 04:49:03.163864 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 04:49:03.163879 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 04:49:03.163892 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.829545
I0607 04:49:03.163904 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.736842
I0607 04:49:03.163923 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.67319 (* 0.3 = 0.501957 loss)
I0607 04:49:03.163938 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.583632 (* 0.3 = 0.17509 loss)
I0607 04:49:03.163952 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 1.15986 (* 0.0272727 = 0.0316326 loss)
I0607 04:49:03.163967 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.19452 (* 0.0272727 = 0.0325779 loss)
I0607 04:49:03.163981 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.86713 (* 0.0272727 = 0.0509216 loss)
I0607 04:49:03.163995 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.7762 (* 0.0272727 = 0.0484417 loss)
I0607 04:49:03.164011 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.89013 (* 0.0272727 = 0.0515489 loss)
I0607 04:49:03.164024 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.36536 (* 0.0272727 = 0.0372372 loss)
I0607 04:49:03.164038 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 1.63919 (* 0.0272727 = 0.0447053 loss)
I0607 04:49:03.164053 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 1.23909 (* 0.0272727 = 0.0337932 loss)
I0607 04:49:03.164067 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.956555 (* 0.0272727 = 0.0260879 loss)
I0607 04:49:03.164083 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 1.06869 (* 0.0272727 = 0.0291461 loss)
I0607 04:49:03.164105 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.242758 (* 0.0272727 = 0.00662067 loss)
I0607 04:49:03.164120 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.447787 (* 0.0272727 = 0.0122124 loss)
I0607 04:49:03.164157 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.670993 (* 0.0272727 = 0.0182998 loss)
I0607 04:49:03.164178 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.0329854 (* 0.0272727 = 0.000899603 loss)
I0607 04:49:03.164192 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.0223037 (* 0.0272727 = 0.000608283 loss)
I0607 04:49:03.164207 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.00265456 (* 0.0272727 = 7.2397e-05 loss)
I0607 04:49:03.164222 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.0052441 (* 0.0272727 = 0.000143021 loss)
I0607 04:49:03.164237 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.0022141 (* 0.0272727 = 6.03845e-05 loss)
I0607 04:49:03.164252 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.000298238 (* 0.0272727 = 8.13377e-06 loss)
I0607 04:49:03.164266 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.00011599 (* 0.0272727 = 3.16336e-06 loss)
I0607 04:49:03.164280 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 6.32936e-05 (* 0.0272727 = 1.72619e-06 loss)
I0607 04:49:03.164295 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 4.01907e-05 (* 0.0272727 = 1.09611e-06 loss)
I0607 04:49:03.164307 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.526316
I0607 04:49:03.164320 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.875
I0607 04:49:03.164333 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.875
I0607 04:49:03.164345 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.75
I0607 04:49:03.164358 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.5
I0607 04:49:03.164371 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.5
I0607 04:49:03.164382 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.625
I0607 04:49:03.164394 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.75
I0607 04:49:03.164407 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.75
I0607 04:49:03.164418 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.75
I0607 04:49:03.164432 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.75
I0607 04:49:03.164443 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.875
I0607 04:49:03.164455 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 1
I0607 04:49:03.164468 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.875
I0607 04:49:03.164479 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 1
I0607 04:49:03.164491 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 04:49:03.164504 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 04:49:03.164515 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 04:49:03.164527 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 04:49:03.164540 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 04:49:03.164551 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 04:49:03.164563 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 04:49:03.164578 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 04:49:03.164592 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.829545
I0607 04:49:03.164603 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.807018
I0607 04:49:03.164618 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.39602 (* 0.3 = 0.418805 loss)
I0607 04:49:03.164634 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.502768 (* 0.3 = 0.15083 loss)
I0607 04:49:03.164649 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.286115 (* 0.0272727 = 0.00780313 loss)
I0607 04:49:03.164664 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.807628 (* 0.0272727 = 0.0220262 loss)
I0607 04:49:03.164690 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 1.55284 (* 0.0272727 = 0.0423503 loss)
I0607 04:49:03.164705 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.67397 (* 0.0272727 = 0.0456538 loss)
I0607 04:49:03.164719 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.64125 (* 0.0272727 = 0.0447614 loss)
I0607 04:49:03.164733 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.46237 (* 0.0272727 = 0.0398828 loss)
I0607 04:49:03.164748 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 1.25643 (* 0.0272727 = 0.0342662 loss)
I0607 04:49:03.164762 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.889838 (* 0.0272727 = 0.0242683 loss)
I0607 04:49:03.164777 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.571013 (* 0.0272727 = 0.0155731 loss)
I0607 04:49:03.164790 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.6924 (* 0.0272727 = 0.0188836 loss)
I0607 04:49:03.164805 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.372826 (* 0.0272727 = 0.010168 loss)
I0607 04:49:03.164819 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.300812 (* 0.0272727 = 0.00820395 loss)
I0607 04:49:03.164834 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.492129 (* 0.0272727 = 0.0134217 loss)
I0607 04:49:03.164849 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.111025 (* 0.0272727 = 0.00302796 loss)
I0607 04:49:03.164863 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.0533052 (* 0.0272727 = 0.00145378 loss)
I0607 04:49:03.164878 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.0140352 (* 0.0272727 = 0.000382778 loss)
I0607 04:49:03.164892 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.0041608 (* 0.0272727 = 0.000113476 loss)
I0607 04:49:03.164907 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.000998796 (* 0.0272727 = 2.72399e-05 loss)
I0607 04:49:03.164921 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.00017952 (* 0.0272727 = 4.89601e-06 loss)
I0607 04:49:03.164939 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.000108021 (* 0.0272727 = 2.94602e-06 loss)
I0607 04:49:03.164953 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 5.56622e-05 (* 0.0272727 = 1.51806e-06 loss)
I0607 04:49:03.164968 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 3.90965e-05 (* 0.0272727 = 1.06627e-06 loss)
I0607 04:49:03.164980 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.754386
I0607 04:49:03.164994 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 04:49:03.165005 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.875
I0607 04:49:03.165017 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.875
I0607 04:49:03.165030 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.875
I0607 04:49:03.165042 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.5
I0607 04:49:03.165055 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.875
I0607 04:49:03.165066 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.875
I0607 04:49:03.165078 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.875
I0607 04:49:03.165091 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.875
I0607 04:49:03.165102 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 1
I0607 04:49:03.165114 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.875
I0607 04:49:03.165141 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 1
I0607 04:49:03.165154 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 0.875
I0607 04:49:03.165166 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 04:49:03.165179 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 04:49:03.165191 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 04:49:03.165215 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 04:49:03.165228 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 04:49:03.165241 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 04:49:03.165252 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 04:49:03.165264 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 04:49:03.165288 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 04:49:03.165310 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.903409
I0607 04:49:03.165338 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.947368
I0607 04:49:03.165355 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.7855 (* 1 = 0.7855 loss)
I0607 04:49:03.165369 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.301631 (* 1 = 0.301631 loss)
I0607 04:49:03.165385 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.0274126 (* 0.0909091 = 0.00249205 loss)
I0607 04:49:03.165400 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.202284 (* 0.0909091 = 0.0183894 loss)
I0607 04:49:03.165415 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.920491 (* 0.0909091 = 0.083681 loss)
I0607 04:49:03.165428 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.672703 (* 0.0909091 = 0.0611548 loss)
I0607 04:49:03.165442 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 1.03855 (* 0.0909091 = 0.0944136 loss)
I0607 04:49:03.165457 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.389678 (* 0.0909091 = 0.0354253 loss)
I0607 04:49:03.165467 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.773227 (* 0.0909091 = 0.0702934 loss)
I0607 04:49:03.165477 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.384229 (* 0.0909091 = 0.03493 loss)
I0607 04:49:03.165493 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.297073 (* 0.0909091 = 0.0270066 loss)
I0607 04:49:03.165508 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.152042 (* 0.0909091 = 0.013822 loss)
I0607 04:49:03.165521 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.19925 (* 0.0909091 = 0.0181137 loss)
I0607 04:49:03.165535 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.264589 (* 0.0909091 = 0.0240536 loss)
I0607 04:49:03.165549 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.24783 (* 0.0909091 = 0.02253 loss)
I0607 04:49:03.165572 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.0289196 (* 0.0909091 = 0.00262905 loss)
I0607 04:49:03.165587 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.0151413 (* 0.0909091 = 0.00137648 loss)
I0607 04:49:03.165601 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.00514795 (* 0.0909091 = 0.000467996 loss)
I0607 04:49:03.165616 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.00154147 (* 0.0909091 = 0.000140134 loss)
I0607 04:49:03.165640 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.00123649 (* 0.0909091 = 0.000112408 loss)
I0607 04:49:03.165655 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.000385983 (* 0.0909091 = 3.50894e-05 loss)
I0607 04:49:03.165669 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.000153807 (* 0.0909091 = 1.39824e-05 loss)
I0607 04:49:03.165684 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.000117187 (* 0.0909091 = 1.06534e-05 loss)
I0607 04:49:03.165699 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 3.39615e-05 (* 0.0909091 = 3.08741e-06 loss)
I0607 04:49:03.165710 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.5
I0607 04:49:03.165722 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.5
I0607 04:49:03.165746 32403 solver.cpp:245] Train net output #149: total_confidence = 0.379022
I0607 04:49:03.165760 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.425332
I0607 04:49:03.165779 32403 sgd_solver.cpp:106] Iteration 17000, lr = 0.001
I0607 04:50:20.681380 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.6962 > 30) by scale factor 0.946487
I0607 04:52:53.580606 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.8103 > 30) by scale factor 0.861814
I0607 04:54:06.834758 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 38.8338 > 30) by scale factor 0.772523
I0607 04:55:17.088270 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.5159 > 30) by scale factor 0.799661
I0607 04:55:18.633997 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 41.4406 > 30) by scale factor 0.723927
I0607 04:55:29.100119 32403 solver.cpp:229] Iteration 17500, loss = 3.95112
I0607 04:55:29.100193 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.338028
I0607 04:55:29.100213 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.5
I0607 04:55:29.100227 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.625
I0607 04:55:29.100240 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.375
I0607 04:55:29.100255 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.375
I0607 04:55:29.100267 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.25
I0607 04:55:29.100281 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.5
I0607 04:55:29.100293 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.5
I0607 04:55:29.100306 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.5
I0607 04:55:29.100318 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.625
I0607 04:55:29.100332 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.625
I0607 04:55:29.100345 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.625
I0607 04:55:29.100358 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 04:55:29.100371 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.875
I0607 04:55:29.100383 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.875
I0607 04:55:29.100396 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 0.875
I0607 04:55:29.100409 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 0.875
I0607 04:55:29.100420 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 0.875
I0607 04:55:29.100432 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 04:55:29.100445 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 04:55:29.100457 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 04:55:29.100469 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 04:55:29.100481 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 04:55:29.100493 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.715909
I0607 04:55:29.100507 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.56338
I0607 04:55:29.100523 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 2.09279 (* 0.3 = 0.627838 loss)
I0607 04:55:29.100538 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.914823 (* 0.3 = 0.274447 loss)
I0607 04:55:29.100553 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 1.49197 (* 0.0272727 = 0.0406901 loss)
I0607 04:55:29.100567 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.36674 (* 0.0272727 = 0.0372748 loss)
I0607 04:55:29.100582 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 2.09076 (* 0.0272727 = 0.0570206 loss)
I0607 04:55:29.100596 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 2.58088 (* 0.0272727 = 0.0703877 loss)
I0607 04:55:29.100610 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 2.42971 (* 0.0272727 = 0.0662649 loss)
I0607 04:55:29.100625 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.70732 (* 0.0272727 = 0.0465632 loss)
I0607 04:55:29.100638 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 1.88874 (* 0.0272727 = 0.0515112 loss)
I0607 04:55:29.100652 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 1.44999 (* 0.0272727 = 0.0395451 loss)
I0607 04:55:29.100666 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 1.11614 (* 0.0272727 = 0.0304402 loss)
I0607 04:55:29.100680 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 1.35095 (* 0.0272727 = 0.0368441 loss)
I0607 04:55:29.100694 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 1.0691 (* 0.0272727 = 0.0291574 loss)
I0607 04:55:29.100751 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.47225 (* 0.0272727 = 0.0128795 loss)
I0607 04:55:29.100770 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.892487 (* 0.0272727 = 0.0243406 loss)
I0607 04:55:29.100785 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.837323 (* 0.0272727 = 0.0228361 loss)
I0607 04:55:29.100800 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.441858 (* 0.0272727 = 0.0120507 loss)
I0607 04:55:29.100822 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.465541 (* 0.0272727 = 0.0126966 loss)
I0607 04:55:29.100836 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.634624 (* 0.0272727 = 0.0173079 loss)
I0607 04:55:29.100852 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.00439115 (* 0.0272727 = 0.000119759 loss)
I0607 04:55:29.100865 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.00353098 (* 0.0272727 = 9.62993e-05 loss)
I0607 04:55:29.100884 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.00241977 (* 0.0272727 = 6.59938e-05 loss)
I0607 04:55:29.100899 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.000885456 (* 0.0272727 = 2.41488e-05 loss)
I0607 04:55:29.100914 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.00156286 (* 0.0272727 = 4.26236e-05 loss)
I0607 04:55:29.100926 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.56338
I0607 04:55:29.100939 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.875
I0607 04:55:29.100951 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.875
I0607 04:55:29.100963 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.75
I0607 04:55:29.100981 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.375
I0607 04:55:29.100997 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.5
I0607 04:55:29.101009 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.5
I0607 04:55:29.101022 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.5
I0607 04:55:29.101043 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.5
I0607 04:55:29.101055 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.625
I0607 04:55:29.101068 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.625
I0607 04:55:29.101079 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.625
I0607 04:55:29.101091 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 1
I0607 04:55:29.101104 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.875
I0607 04:55:29.101116 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.875
I0607 04:55:29.101143 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 0.875
I0607 04:55:29.101156 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 0.875
I0607 04:55:29.101169 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 0.875
I0607 04:55:29.101181 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 04:55:29.101193 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 04:55:29.101204 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 04:55:29.101217 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 04:55:29.101235 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 04:55:29.101248 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.818182
I0607 04:55:29.101259 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.732394
I0607 04:55:29.101274 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.57702 (* 0.3 = 0.473107 loss)
I0607 04:55:29.101291 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.657951 (* 0.3 = 0.197385 loss)
I0607 04:55:29.101318 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.999029 (* 0.0272727 = 0.0272463 loss)
I0607 04:55:29.101335 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.832878 (* 0.0272727 = 0.0227149 loss)
I0607 04:55:29.101348 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.895286 (* 0.0272727 = 0.0244169 loss)
I0607 04:55:29.101363 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.55612 (* 0.0272727 = 0.0424398 loss)
I0607 04:55:29.101377 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.53266 (* 0.0272727 = 0.0417998 loss)
I0607 04:55:29.101392 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.48848 (* 0.0272727 = 0.040595 loss)
I0607 04:55:29.101405 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 2.11007 (* 0.0272727 = 0.0575473 loss)
I0607 04:55:29.101419 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 1.17634 (* 0.0272727 = 0.0320821 loss)
I0607 04:55:29.101433 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 1.51466 (* 0.0272727 = 0.041309 loss)
I0607 04:55:29.101447 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 1.34511 (* 0.0272727 = 0.0366849 loss)
I0607 04:55:29.101461 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 1.48325 (* 0.0272727 = 0.0404523 loss)
I0607 04:55:29.101475 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.259746 (* 0.0272727 = 0.00708398 loss)
I0607 04:55:29.101490 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.830272 (* 0.0272727 = 0.0226438 loss)
I0607 04:55:29.101505 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.627887 (* 0.0272727 = 0.0171242 loss)
I0607 04:55:29.101523 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.349252 (* 0.0272727 = 0.00952505 loss)
I0607 04:55:29.101537 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.457145 (* 0.0272727 = 0.0124676 loss)
I0607 04:55:29.101552 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.6832 (* 0.0272727 = 0.0186327 loss)
I0607 04:55:29.101574 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.00680572 (* 0.0272727 = 0.00018561 loss)
I0607 04:55:29.101588 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.0024442 (* 0.0272727 = 6.66599e-05 loss)
I0607 04:55:29.101603 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.00247274 (* 0.0272727 = 6.74383e-05 loss)
I0607 04:55:29.101618 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.000755861 (* 0.0272727 = 2.06144e-05 loss)
I0607 04:55:29.101632 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 0.00121702 (* 0.0272727 = 3.31914e-05 loss)
I0607 04:55:29.101644 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.690141
I0607 04:55:29.101657 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 0.875
I0607 04:55:29.101670 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.875
I0607 04:55:29.101681 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.875
I0607 04:55:29.101693 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.875
I0607 04:55:29.101706 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.875
I0607 04:55:29.101718 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.875
I0607 04:55:29.101730 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.625
I0607 04:55:29.101742 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.625
I0607 04:55:29.101754 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.625
I0607 04:55:29.101766 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.625
I0607 04:55:29.101778 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.75
I0607 04:55:29.101790 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.875
I0607 04:55:29.101802 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 0.875
I0607 04:55:29.101827 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 0.875
I0607 04:55:29.101840 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 0.875
I0607 04:55:29.101853 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 0.875
I0607 04:55:29.101866 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 0.875
I0607 04:55:29.101877 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 04:55:29.101889 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 04:55:29.101898 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 04:55:29.101907 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 04:55:29.101919 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 04:55:29.101932 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.869318
I0607 04:55:29.101943 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.859155
I0607 04:55:29.101958 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.99849 (* 1 = 0.99849 loss)
I0607 04:55:29.101972 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.419338 (* 1 = 0.419338 loss)
I0607 04:55:29.101987 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.500921 (* 0.0909091 = 0.0455383 loss)
I0607 04:55:29.102001 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.235951 (* 0.0909091 = 0.0214501 loss)
I0607 04:55:29.102016 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.19464 (* 0.0909091 = 0.0176946 loss)
I0607 04:55:29.102030 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.323939 (* 0.0909091 = 0.029449 loss)
I0607 04:55:29.102047 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.642724 (* 0.0909091 = 0.0584295 loss)
I0607 04:55:29.102062 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.735601 (* 0.0909091 = 0.0668728 loss)
I0607 04:55:29.102077 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 1.51446 (* 0.0909091 = 0.137678 loss)
I0607 04:55:29.102090 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 1.08036 (* 0.0909091 = 0.0982143 loss)
I0607 04:55:29.102105 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 1.01955 (* 0.0909091 = 0.092686 loss)
I0607 04:55:29.102119 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 1.00314 (* 0.0909091 = 0.0911944 loss)
I0607 04:55:29.102133 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.814921 (* 0.0909091 = 0.0740838 loss)
I0607 04:55:29.102147 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.349463 (* 0.0909091 = 0.0317694 loss)
I0607 04:55:29.102161 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.591264 (* 0.0909091 = 0.0537513 loss)
I0607 04:55:29.102175 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.571327 (* 0.0909091 = 0.0519388 loss)
I0607 04:55:29.102190 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.460188 (* 0.0909091 = 0.0418352 loss)
I0607 04:55:29.102203 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.422138 (* 0.0909091 = 0.0383762 loss)
I0607 04:55:29.102217 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.504773 (* 0.0909091 = 0.0458884 loss)
I0607 04:55:29.102232 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.00860102 (* 0.0909091 = 0.000781911 loss)
I0607 04:55:29.102246 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.00169339 (* 0.0909091 = 0.000153945 loss)
I0607 04:55:29.102262 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.00128121 (* 0.0909091 = 0.000116474 loss)
I0607 04:55:29.102275 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.00077064 (* 0.0909091 = 7.00582e-05 loss)
I0607 04:55:29.102289 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 0.000296118 (* 0.0909091 = 2.69198e-05 loss)
I0607 04:55:29.102313 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.375
I0607 04:55:29.102326 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.375
I0607 04:55:29.102339 32403 solver.cpp:245] Train net output #149: total_confidence = 0.362681
I0607 04:55:29.102350 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.300727
I0607 04:55:29.102363 32403 sgd_solver.cpp:106] Iteration 17500, lr = 0.001
I0607 04:57:23.745394 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.0665 > 30) by scale factor 0.88063
I0607 05:00:07.415630 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.9103 > 30) by scale factor 0.97055
I0607 05:01:55.119011 32403 solver.cpp:229] Iteration 18000, loss = 3.92854
I0607 05:01:55.119134 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.458333
I0607 05:01:55.119155 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.75
I0607 05:01:55.119169 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.5
I0607 05:01:55.119182 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.5
I0607 05:01:55.119195 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.375
I0607 05:01:55.119209 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.375
I0607 05:01:55.119221 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.25
I0607 05:01:55.119233 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.5
I0607 05:01:55.119246 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.375
I0607 05:01:55.119259 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.625
I0607 05:01:55.119272 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.75
I0607 05:01:55.119285 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.75
I0607 05:01:55.119297 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.75
I0607 05:01:55.119309 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.875
I0607 05:01:55.119321 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.875
I0607 05:01:55.119334 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 05:01:55.119346 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 05:01:55.119359 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 05:01:55.119370 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 05:01:55.119382 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 05:01:55.119393 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 05:01:55.119405 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 05:01:55.119417 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 05:01:55.119429 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.761364
I0607 05:01:55.119442 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.75
I0607 05:01:55.119458 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.71458 (* 0.3 = 0.514374 loss)
I0607 05:01:55.119473 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.774608 (* 0.3 = 0.232382 loss)
I0607 05:01:55.119488 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.670072 (* 0.0272727 = 0.0182747 loss)
I0607 05:01:55.119503 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.67783 (* 0.0272727 = 0.0457589 loss)
I0607 05:01:55.119518 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.31396 (* 0.0272727 = 0.0358353 loss)
I0607 05:01:55.119531 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.63801 (* 0.0272727 = 0.0446731 loss)
I0607 05:01:55.119545 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.80008 (* 0.0272727 = 0.0490931 loss)
I0607 05:01:55.119560 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 2.07398 (* 0.0272727 = 0.0565632 loss)
I0607 05:01:55.119573 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 1.47478 (* 0.0272727 = 0.0402214 loss)
I0607 05:01:55.119587 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 2.22713 (* 0.0272727 = 0.0607399 loss)
I0607 05:01:55.119601 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 1.07101 (* 0.0272727 = 0.0292094 loss)
I0607 05:01:55.119616 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 1.40928 (* 0.0272727 = 0.038435 loss)
I0607 05:01:55.119629 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.896875 (* 0.0272727 = 0.0244602 loss)
I0607 05:01:55.119643 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.792419 (* 0.0272727 = 0.0216114 loss)
I0607 05:01:55.119676 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.744588 (* 0.0272727 = 0.020307 loss)
I0607 05:01:55.119691 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.651145 (* 0.0272727 = 0.0177585 loss)
I0607 05:01:55.119706 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.280669 (* 0.0272727 = 0.00765462 loss)
I0607 05:01:55.119720 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.102987 (* 0.0272727 = 0.00280873 loss)
I0607 05:01:55.119735 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.0435726 (* 0.0272727 = 0.00118834 loss)
I0607 05:01:55.119750 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.0324916 (* 0.0272727 = 0.000886135 loss)
I0607 05:01:55.119763 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.020184 (* 0.0272727 = 0.000550472 loss)
I0607 05:01:55.119778 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.0384058 (* 0.0272727 = 0.00104743 loss)
I0607 05:01:55.119792 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.0158571 (* 0.0272727 = 0.000432467 loss)
I0607 05:01:55.119807 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.0163674 (* 0.0272727 = 0.000446382 loss)
I0607 05:01:55.119819 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.569444
I0607 05:01:55.119832 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 1
I0607 05:01:55.119844 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.75
I0607 05:01:55.119856 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.625
I0607 05:01:55.119868 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.625
I0607 05:01:55.119884 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.375
I0607 05:01:55.119895 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.375
I0607 05:01:55.119907 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.375
I0607 05:01:55.119920 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.625
I0607 05:01:55.119931 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.625
I0607 05:01:55.119943 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.75
I0607 05:01:55.119956 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.875
I0607 05:01:55.119968 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.75
I0607 05:01:55.119981 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.75
I0607 05:01:55.119992 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.875
I0607 05:01:55.120004 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 05:01:55.120017 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 05:01:55.120028 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 05:01:55.120039 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 05:01:55.120051 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 05:01:55.120062 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 05:01:55.120074 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 05:01:55.120085 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 05:01:55.120097 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.818182
I0607 05:01:55.120110 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.861111
I0607 05:01:55.120123 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.40535 (* 0.3 = 0.421604 loss)
I0607 05:01:55.120136 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.605831 (* 0.3 = 0.181749 loss)
I0607 05:01:55.120154 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.287051 (* 0.0272727 = 0.00782865 loss)
I0607 05:01:55.120169 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.787917 (* 0.0272727 = 0.0214886 loss)
I0607 05:01:55.120195 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 1.29672 (* 0.0272727 = 0.035365 loss)
I0607 05:01:55.120210 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 0.962032 (* 0.0272727 = 0.0262372 loss)
I0607 05:01:55.120224 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.37814 (* 0.0272727 = 0.0375855 loss)
I0607 05:01:55.120239 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.70365 (* 0.0272727 = 0.0464632 loss)
I0607 05:01:55.120252 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 1.49309 (* 0.0272727 = 0.0407206 loss)
I0607 05:01:55.120266 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 1.9428 (* 0.0272727 = 0.0529855 loss)
I0607 05:01:55.120280 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 1.22208 (* 0.0272727 = 0.0333296 loss)
I0607 05:01:55.120295 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 1.26011 (* 0.0272727 = 0.0343667 loss)
I0607 05:01:55.120308 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.976186 (* 0.0272727 = 0.0266232 loss)
I0607 05:01:55.120322 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.772482 (* 0.0272727 = 0.0210677 loss)
I0607 05:01:55.120337 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.707941 (* 0.0272727 = 0.0193075 loss)
I0607 05:01:55.120350 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.342531 (* 0.0272727 = 0.00934175 loss)
I0607 05:01:55.120364 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.165604 (* 0.0272727 = 0.00451647 loss)
I0607 05:01:55.120378 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.0322911 (* 0.0272727 = 0.000880665 loss)
I0607 05:01:55.120393 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.0112286 (* 0.0272727 = 0.000306234 loss)
I0607 05:01:55.120407 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.00438316 (* 0.0272727 = 0.000119541 loss)
I0607 05:01:55.120421 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.000645616 (* 0.0272727 = 1.76077e-05 loss)
I0607 05:01:55.120436 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.00051753 (* 0.0272727 = 1.41145e-05 loss)
I0607 05:01:55.120450 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.00040559 (* 0.0272727 = 1.10615e-05 loss)
I0607 05:01:55.120465 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 0.000352596 (* 0.0272727 = 9.61626e-06 loss)
I0607 05:01:55.120477 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.708333
I0607 05:01:55.120489 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 05:01:55.120501 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.875
I0607 05:01:55.120513 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.875
I0607 05:01:55.120525 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.75
I0607 05:01:55.120537 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.875
I0607 05:01:55.120548 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.875
I0607 05:01:55.120560 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.875
I0607 05:01:55.120573 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.625
I0607 05:01:55.120584 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.625
I0607 05:01:55.120595 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.75
I0607 05:01:55.120607 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.875
I0607 05:01:55.120620 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.75
I0607 05:01:55.120630 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 0.75
I0607 05:01:55.120642 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 05:01:55.120654 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 0.875
I0607 05:01:55.120666 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 05:01:55.120683 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 05:01:55.120692 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 05:01:55.120705 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 05:01:55.120718 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 05:01:55.120729 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 05:01:55.120741 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 05:01:55.120753 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.869318
I0607 05:01:55.120764 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.930556
I0607 05:01:55.120779 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.810454 (* 1 = 0.810454 loss)
I0607 05:01:55.120792 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.37637 (* 1 = 0.37637 loss)
I0607 05:01:55.120807 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.0332289 (* 0.0909091 = 0.0030208 loss)
I0607 05:01:55.120821 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.376304 (* 0.0909091 = 0.0342094 loss)
I0607 05:01:55.120836 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.360095 (* 0.0909091 = 0.0327359 loss)
I0607 05:01:55.120849 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.569269 (* 0.0909091 = 0.0517518 loss)
I0607 05:01:55.120863 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.596603 (* 0.0909091 = 0.0542366 loss)
I0607 05:01:55.120877 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.576738 (* 0.0909091 = 0.0524308 loss)
I0607 05:01:55.120892 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.671485 (* 0.0909091 = 0.0610441 loss)
I0607 05:01:55.120905 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 1.88329 (* 0.0909091 = 0.171208 loss)
I0607 05:01:55.120919 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 1.13402 (* 0.0909091 = 0.103092 loss)
I0607 05:01:55.120936 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.915475 (* 0.0909091 = 0.083225 loss)
I0607 05:01:55.120950 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.761226 (* 0.0909091 = 0.0692023 loss)
I0607 05:01:55.120965 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.389914 (* 0.0909091 = 0.0354467 loss)
I0607 05:01:55.120978 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.682831 (* 0.0909091 = 0.0620756 loss)
I0607 05:01:55.120992 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.268627 (* 0.0909091 = 0.0244206 loss)
I0607 05:01:55.121006 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.208443 (* 0.0909091 = 0.0189493 loss)
I0607 05:01:55.121021 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.0713616 (* 0.0909091 = 0.00648742 loss)
I0607 05:01:55.121034 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.0179212 (* 0.0909091 = 0.0016292 loss)
I0607 05:01:55.121049 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.0119939 (* 0.0909091 = 0.00109035 loss)
I0607 05:01:55.121063 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.0140374 (* 0.0909091 = 0.00127613 loss)
I0607 05:01:55.121078 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.00461939 (* 0.0909091 = 0.000419944 loss)
I0607 05:01:55.121091 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.00313347 (* 0.0909091 = 0.000284861 loss)
I0607 05:01:55.121105 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 0.00112512 (* 0.0909091 = 0.000102283 loss)
I0607 05:01:55.121129 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.375
I0607 05:01:55.121145 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.375
I0607 05:01:55.121167 32403 solver.cpp:245] Train net output #149: total_confidence = 0.238825
I0607 05:01:55.121181 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.226598
I0607 05:01:55.121193 32403 sgd_solver.cpp:106] Iteration 18000, lr = 0.001
I0607 05:05:05.346926 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.7912 > 30) by scale factor 0.943658
I0607 05:05:45.465961 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 36.8459 > 30) by scale factor 0.814202
I0607 05:06:23.278255 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.8029 > 30) by scale factor 0.914554
I0607 05:06:33.305114 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 46.7202 > 30) by scale factor 0.64212
I0607 05:07:32.719344 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.4578 > 30) by scale factor 0.98497
I0607 05:07:55.089129 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 38.8041 > 30) by scale factor 0.773114
I0607 05:08:19.017480 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.1318 > 30) by scale factor 0.963645
I0607 05:08:20.970680 32403 solver.cpp:229] Iteration 18500, loss = 3.93066
I0607 05:08:20.970752 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.584906
I0607 05:08:20.970772 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.75
I0607 05:08:20.970785 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.75
I0607 05:08:20.970798 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.5
I0607 05:08:20.970811 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.75
I0607 05:08:20.970824 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.375
I0607 05:08:20.970837 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.5
I0607 05:08:20.970850 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.75
I0607 05:08:20.970862 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 1
I0607 05:08:20.970876 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 1
I0607 05:08:20.970890 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.875
I0607 05:08:20.970902 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.875
I0607 05:08:20.970914 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 05:08:20.970928 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 1
I0607 05:08:20.970940 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 1
I0607 05:08:20.970952 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 05:08:20.970964 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 05:08:20.970976 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 05:08:20.970988 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 05:08:20.971000 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 05:08:20.971014 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 05:08:20.971025 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 05:08:20.971037 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 05:08:20.971050 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.863636
I0607 05:08:20.971062 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.849057
I0607 05:08:20.971078 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.15548 (* 0.3 = 0.346644 loss)
I0607 05:08:20.971093 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.380662 (* 0.3 = 0.114198 loss)
I0607 05:08:20.971108 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.644504 (* 0.0272727 = 0.0175774 loss)
I0607 05:08:20.971123 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.3154 (* 0.0272727 = 0.0358746 loss)
I0607 05:08:20.971138 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.14839 (* 0.0272727 = 0.0313197 loss)
I0607 05:08:20.971153 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 0.996346 (* 0.0272727 = 0.0271731 loss)
I0607 05:08:20.971166 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.70287 (* 0.0272727 = 0.046442 loss)
I0607 05:08:20.971180 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.08512 (* 0.0272727 = 0.0295941 loss)
I0607 05:08:20.971194 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 1.23545 (* 0.0272727 = 0.0336942 loss)
I0607 05:08:20.971210 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.208393 (* 0.0272727 = 0.00568345 loss)
I0607 05:08:20.971223 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.205127 (* 0.0272727 = 0.00559438 loss)
I0607 05:08:20.971238 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.306525 (* 0.0272727 = 0.00835977 loss)
I0607 05:08:20.971252 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.349532 (* 0.0272727 = 0.0095327 loss)
I0607 05:08:20.971267 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.318723 (* 0.0272727 = 0.00869246 loss)
I0607 05:08:20.971318 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.0638125 (* 0.0272727 = 0.00174034 loss)
I0607 05:08:20.971334 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.0420586 (* 0.0272727 = 0.00114705 loss)
I0607 05:08:20.971349 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.010843 (* 0.0272727 = 0.000295719 loss)
I0607 05:08:20.971364 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.00357409 (* 0.0272727 = 9.74751e-05 loss)
I0607 05:08:20.971379 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.000449703 (* 0.0272727 = 1.22646e-05 loss)
I0607 05:08:20.971392 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.000354492 (* 0.0272727 = 9.66795e-06 loss)
I0607 05:08:20.971407 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.000129946 (* 0.0272727 = 3.54397e-06 loss)
I0607 05:08:20.971421 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 1.21897e-05 (* 0.0272727 = 3.32447e-07 loss)
I0607 05:08:20.971436 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 5.43902e-06 (* 0.0272727 = 1.48337e-07 loss)
I0607 05:08:20.971451 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 1.29641e-06 (* 0.0272727 = 3.53565e-08 loss)
I0607 05:08:20.971462 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.716981
I0607 05:08:20.971475 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.875
I0607 05:08:20.971487 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.75
I0607 05:08:20.971500 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.875
I0607 05:08:20.971513 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.5
I0607 05:08:20.971524 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.625
I0607 05:08:20.971536 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.625
I0607 05:08:20.971549 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.625
I0607 05:08:20.971561 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 1
I0607 05:08:20.971573 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.875
I0607 05:08:20.971586 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.875
I0607 05:08:20.971598 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.875
I0607 05:08:20.971611 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.875
I0607 05:08:20.971622 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 1
I0607 05:08:20.971634 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 1
I0607 05:08:20.971647 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 05:08:20.971658 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 05:08:20.971670 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 05:08:20.971683 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 05:08:20.971698 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 05:08:20.971710 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 05:08:20.971722 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 05:08:20.971735 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 05:08:20.971746 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.903409
I0607 05:08:20.971758 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.886792
I0607 05:08:20.971773 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 0.874386 (* 0.3 = 0.262316 loss)
I0607 05:08:20.971787 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.312235 (* 0.3 = 0.0936706 loss)
I0607 05:08:20.971802 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.35632 (* 0.0272727 = 0.00971783 loss)
I0607 05:08:20.971828 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.856332 (* 0.0272727 = 0.0233545 loss)
I0607 05:08:20.971843 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.411803 (* 0.0272727 = 0.011231 loss)
I0607 05:08:20.971858 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.02269 (* 0.0272727 = 0.0278917 loss)
I0607 05:08:20.971873 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.03367 (* 0.0272727 = 0.028191 loss)
I0607 05:08:20.971886 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 0.877641 (* 0.0272727 = 0.0239357 loss)
I0607 05:08:20.971901 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 1.03313 (* 0.0272727 = 0.0281763 loss)
I0607 05:08:20.971915 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.388978 (* 0.0272727 = 0.0106085 loss)
I0607 05:08:20.971930 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.406278 (* 0.0272727 = 0.0110803 loss)
I0607 05:08:20.971943 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.593898 (* 0.0272727 = 0.0161972 loss)
I0607 05:08:20.971958 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.353233 (* 0.0272727 = 0.00963364 loss)
I0607 05:08:20.971973 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.475783 (* 0.0272727 = 0.0129759 loss)
I0607 05:08:20.971987 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.0170447 (* 0.0272727 = 0.000464855 loss)
I0607 05:08:20.972002 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.00898022 (* 0.0272727 = 0.000244915 loss)
I0607 05:08:20.972017 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.00413831 (* 0.0272727 = 0.000112863 loss)
I0607 05:08:20.972030 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.00565031 (* 0.0272727 = 0.000154099 loss)
I0607 05:08:20.972045 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.00255349 (* 0.0272727 = 6.96405e-05 loss)
I0607 05:08:20.972059 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.00172992 (* 0.0272727 = 4.71797e-05 loss)
I0607 05:08:20.972074 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.000754243 (* 0.0272727 = 2.05703e-05 loss)
I0607 05:08:20.972089 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.000620637 (* 0.0272727 = 1.69265e-05 loss)
I0607 05:08:20.972102 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.000132149 (* 0.0272727 = 3.60406e-06 loss)
I0607 05:08:20.972116 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 5.21296e-05 (* 0.0272727 = 1.42172e-06 loss)
I0607 05:08:20.972129 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.90566
I0607 05:08:20.972141 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 05:08:20.972154 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.875
I0607 05:08:20.972167 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 1
I0607 05:08:20.972178 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 1
I0607 05:08:20.972190 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 1
I0607 05:08:20.972203 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.875
I0607 05:08:20.972211 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.875
I0607 05:08:20.972219 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 1
I0607 05:08:20.972226 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.875
I0607 05:08:20.972239 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.875
I0607 05:08:20.972251 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.875
I0607 05:08:20.972264 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.875
I0607 05:08:20.972275 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 1
I0607 05:08:20.972287 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 05:08:20.972300 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 05:08:20.972322 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 05:08:20.972335 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 05:08:20.972347 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 05:08:20.972360 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 05:08:20.972373 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 05:08:20.972384 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 05:08:20.972396 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 05:08:20.972407 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.971591
I0607 05:08:20.972420 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.962264
I0607 05:08:20.972434 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.356043 (* 1 = 0.356043 loss)
I0607 05:08:20.972448 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.117682 (* 1 = 0.117682 loss)
I0607 05:08:20.972463 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.142087 (* 0.0909091 = 0.012917 loss)
I0607 05:08:20.972477 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.32253 (* 0.0909091 = 0.0293209 loss)
I0607 05:08:20.972492 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.0731483 (* 0.0909091 = 0.00664985 loss)
I0607 05:08:20.972506 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.120444 (* 0.0909091 = 0.0109495 loss)
I0607 05:08:20.972520 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.264794 (* 0.0909091 = 0.0240722 loss)
I0607 05:08:20.972535 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.424308 (* 0.0909091 = 0.0385735 loss)
I0607 05:08:20.972549 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.613293 (* 0.0909091 = 0.0557539 loss)
I0607 05:08:20.972563 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.0866484 (* 0.0909091 = 0.00787713 loss)
I0607 05:08:20.972579 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.215863 (* 0.0909091 = 0.0196239 loss)
I0607 05:08:20.972592 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.399363 (* 0.0909091 = 0.0363057 loss)
I0607 05:08:20.972606 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.312952 (* 0.0909091 = 0.0284501 loss)
I0607 05:08:20.972620 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.218592 (* 0.0909091 = 0.019872 loss)
I0607 05:08:20.972635 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.0228468 (* 0.0909091 = 0.00207698 loss)
I0607 05:08:20.972648 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.00522684 (* 0.0909091 = 0.000475167 loss)
I0607 05:08:20.972662 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.00139666 (* 0.0909091 = 0.000126969 loss)
I0607 05:08:20.972676 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.00058577 (* 0.0909091 = 5.32518e-05 loss)
I0607 05:08:20.972692 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.000168243 (* 0.0909091 = 1.52948e-05 loss)
I0607 05:08:20.972705 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 8.46809e-05 (* 0.0909091 = 7.69827e-06 loss)
I0607 05:08:20.972719 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 7.62956e-05 (* 0.0909091 = 6.93597e-06 loss)
I0607 05:08:20.972733 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 2.84928e-05 (* 0.0909091 = 2.59026e-06 loss)
I0607 05:08:20.972751 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 2.35152e-05 (* 0.0909091 = 2.13775e-06 loss)
I0607 05:08:20.972766 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 1.94915e-05 (* 0.0909091 = 1.77196e-06 loss)
I0607 05:08:20.972779 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.75
I0607 05:08:20.972801 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.75
I0607 05:08:20.972815 32403 solver.cpp:245] Train net output #149: total_confidence = 0.617044
I0607 05:08:20.972827 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.603664
I0607 05:08:20.972841 32403 sgd_solver.cpp:106] Iteration 18500, lr = 0.001
I0607 05:10:23.325101 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.1439 > 30) by scale factor 0.995227
I0607 05:10:38.004803 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.9821 > 30) by scale factor 0.938024
I0607 05:12:07.535873 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.775 > 30) by scale factor 0.862688
I0607 05:12:31.517904 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 62.5683 > 30) by scale factor 0.479476
I0607 05:13:15.551697 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 36.5573 > 30) by scale factor 0.820629
I0607 05:13:16.325340 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 45.2424 > 30) by scale factor 0.663094
I0607 05:14:21.208791 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.1649 > 30) by scale factor 0.994534
I0607 05:14:47.153373 32403 solver.cpp:229] Iteration 19000, loss = 3.99758
I0607 05:14:47.153432 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.583333
I0607 05:14:47.153451 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.75
I0607 05:14:47.153465 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.5
I0607 05:14:47.153478 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.625
I0607 05:14:47.153492 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.375
I0607 05:14:47.153506 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.5
I0607 05:14:47.153518 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.5
I0607 05:14:47.153530 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.75
I0607 05:14:47.153543 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.875
I0607 05:14:47.153556 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.875
I0607 05:14:47.153569 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.875
I0607 05:14:47.153583 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.875
I0607 05:14:47.153594 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 1
I0607 05:14:47.153607 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 1
I0607 05:14:47.153620 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 1
I0607 05:14:47.153631 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 05:14:47.153643 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 05:14:47.153656 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 05:14:47.153671 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 05:14:47.153682 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 05:14:47.153695 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 05:14:47.153707 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 05:14:47.153719 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 05:14:47.153733 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.875
I0607 05:14:47.153746 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.8125
I0607 05:14:47.153764 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.44356 (* 0.3 = 0.433069 loss)
I0607 05:14:47.153777 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.421332 (* 0.3 = 0.1264 loss)
I0607 05:14:47.153792 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.423321 (* 0.0272727 = 0.0115451 loss)
I0607 05:14:47.153806 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.66037 (* 0.0272727 = 0.0452828 loss)
I0607 05:14:47.153821 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.1942 (* 0.0272727 = 0.032569 loss)
I0607 05:14:47.153836 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.52671 (* 0.0272727 = 0.0416375 loss)
I0607 05:14:47.153849 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.56639 (* 0.0272727 = 0.0427198 loss)
I0607 05:14:47.153863 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.36932 (* 0.0272727 = 0.0373451 loss)
I0607 05:14:47.153878 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 0.718002 (* 0.0272727 = 0.0195819 loss)
I0607 05:14:47.153892 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.638373 (* 0.0272727 = 0.0174102 loss)
I0607 05:14:47.153906 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.381064 (* 0.0272727 = 0.0103927 loss)
I0607 05:14:47.153921 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.368581 (* 0.0272727 = 0.0100522 loss)
I0607 05:14:47.153935 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.460092 (* 0.0272727 = 0.012548 loss)
I0607 05:14:47.153950 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.009969 (* 0.0272727 = 0.000271882 loss)
I0607 05:14:47.153997 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.00201154 (* 0.0272727 = 5.48603e-05 loss)
I0607 05:14:47.154014 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.00149873 (* 0.0272727 = 4.08745e-05 loss)
I0607 05:14:47.154029 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.000586214 (* 0.0272727 = 1.59877e-05 loss)
I0607 05:14:47.154044 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.000108632 (* 0.0272727 = 2.96268e-06 loss)
I0607 05:14:47.154058 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 9.42959e-05 (* 0.0272727 = 2.57171e-06 loss)
I0607 05:14:47.154073 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 4.57083e-05 (* 0.0272727 = 1.24659e-06 loss)
I0607 05:14:47.154088 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 3.28971e-05 (* 0.0272727 = 8.97194e-07 loss)
I0607 05:14:47.154103 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 1.2711e-05 (* 0.0272727 = 3.46663e-07 loss)
I0607 05:14:47.154116 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 8.01693e-06 (* 0.0272727 = 2.18644e-07 loss)
I0607 05:14:47.154131 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 7.74874e-06 (* 0.0272727 = 2.11329e-07 loss)
I0607 05:14:47.154145 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.708333
I0607 05:14:47.154157 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 1
I0607 05:14:47.154170 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.875
I0607 05:14:47.154181 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 1
I0607 05:14:47.154192 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.625
I0607 05:14:47.154204 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.75
I0607 05:14:47.154217 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.75
I0607 05:14:47.154228 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.75
I0607 05:14:47.154240 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.75
I0607 05:14:47.154252 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.875
I0607 05:14:47.154270 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.875
I0607 05:14:47.154281 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.875
I0607 05:14:47.154294 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 1
I0607 05:14:47.154305 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 1
I0607 05:14:47.154317 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 1
I0607 05:14:47.154333 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 05:14:47.154345 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 05:14:47.154356 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 05:14:47.154367 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 05:14:47.154379 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 05:14:47.154391 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 05:14:47.154407 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 05:14:47.154420 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 05:14:47.154443 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.909091
I0607 05:14:47.154466 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.833333
I0607 05:14:47.154491 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.03808 (* 0.3 = 0.311423 loss)
I0607 05:14:47.154505 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.316477 (* 0.3 = 0.094943 loss)
I0607 05:14:47.154520 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.120357 (* 0.0272727 = 0.00328245 loss)
I0607 05:14:47.154556 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.619574 (* 0.0272727 = 0.0168975 loss)
I0607 05:14:47.154570 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.392696 (* 0.0272727 = 0.0107099 loss)
I0607 05:14:47.154585 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.25358 (* 0.0272727 = 0.0341885 loss)
I0607 05:14:47.154599 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.1115 (* 0.0272727 = 0.0303136 loss)
I0607 05:14:47.154613 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 0.827376 (* 0.0272727 = 0.0225648 loss)
I0607 05:14:47.154628 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.611606 (* 0.0272727 = 0.0166802 loss)
I0607 05:14:47.154641 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.715913 (* 0.0272727 = 0.0195249 loss)
I0607 05:14:47.154655 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.644987 (* 0.0272727 = 0.0175905 loss)
I0607 05:14:47.154670 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.465393 (* 0.0272727 = 0.0126925 loss)
I0607 05:14:47.154683 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.688324 (* 0.0272727 = 0.0187725 loss)
I0607 05:14:47.154698 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.00154382 (* 0.0272727 = 4.2104e-05 loss)
I0607 05:14:47.154712 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.000187912 (* 0.0272727 = 5.12488e-06 loss)
I0607 05:14:47.154731 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 4.25845e-05 (* 0.0272727 = 1.1614e-06 loss)
I0607 05:14:47.154747 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 1.49016e-05 (* 0.0272727 = 4.06406e-07 loss)
I0607 05:14:47.154762 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 8.28516e-06 (* 0.0272727 = 2.25959e-07 loss)
I0607 05:14:47.154775 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 8.12129e-06 (* 0.0272727 = 2.2149e-07 loss)
I0607 05:14:47.154793 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 2.22028e-06 (* 0.0272727 = 6.05531e-08 loss)
I0607 05:14:47.154806 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 1.17719e-06 (* 0.0272727 = 3.21053e-08 loss)
I0607 05:14:47.154821 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 7.45059e-07 (* 0.0272727 = 2.03198e-08 loss)
I0607 05:14:47.154835 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 4.76837e-07 (* 0.0272727 = 1.30047e-08 loss)
I0607 05:14:47.154850 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 8.64268e-07 (* 0.0272727 = 2.35709e-08 loss)
I0607 05:14:47.154861 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.854167
I0607 05:14:47.154875 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 05:14:47.154886 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.875
I0607 05:14:47.154897 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 1
I0607 05:14:47.154909 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.875
I0607 05:14:47.154922 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.875
I0607 05:14:47.154933 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 1
I0607 05:14:47.154945 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 1
I0607 05:14:47.154956 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.75
I0607 05:14:47.154968 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.875
I0607 05:14:47.154980 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 1
I0607 05:14:47.154991 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.875
I0607 05:14:47.155004 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 1
I0607 05:14:47.155015 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 1
I0607 05:14:47.155024 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 05:14:47.155030 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 05:14:47.155053 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 05:14:47.155066 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 05:14:47.155078 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 05:14:47.155089 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 05:14:47.155102 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 05:14:47.155113 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 05:14:47.155125 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 05:14:47.155136 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.954545
I0607 05:14:47.155148 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.916667
I0607 05:14:47.155164 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.503231 (* 1 = 0.503231 loss)
I0607 05:14:47.155177 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.153013 (* 1 = 0.153013 loss)
I0607 05:14:47.155191 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.0229418 (* 0.0909091 = 0.00208562 loss)
I0607 05:14:47.155206 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.639694 (* 0.0909091 = 0.058154 loss)
I0607 05:14:47.155221 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.0839461 (* 0.0909091 = 0.00763147 loss)
I0607 05:14:47.155236 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.71048 (* 0.0909091 = 0.0645891 loss)
I0607 05:14:47.155249 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.619273 (* 0.0909091 = 0.0562976 loss)
I0607 05:14:47.155263 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.304716 (* 0.0909091 = 0.0277014 loss)
I0607 05:14:47.155278 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.119234 (* 0.0909091 = 0.0108394 loss)
I0607 05:14:47.155292 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.351197 (* 0.0909091 = 0.031927 loss)
I0607 05:14:47.155306 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.316182 (* 0.0909091 = 0.0287438 loss)
I0607 05:14:47.155320 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.130959 (* 0.0909091 = 0.0119054 loss)
I0607 05:14:47.155334 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.317143 (* 0.0909091 = 0.0288312 loss)
I0607 05:14:47.155349 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.0466586 (* 0.0909091 = 0.00424169 loss)
I0607 05:14:47.155364 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.00789804 (* 0.0909091 = 0.000718004 loss)
I0607 05:14:47.155377 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.00102703 (* 0.0909091 = 9.33664e-05 loss)
I0607 05:14:47.155391 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.000466407 (* 0.0909091 = 4.24006e-05 loss)
I0607 05:14:47.155406 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.00016456 (* 0.0909091 = 1.496e-05 loss)
I0607 05:14:47.155421 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.000165773 (* 0.0909091 = 1.50703e-05 loss)
I0607 05:14:47.155434 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.000153842 (* 0.0909091 = 1.39856e-05 loss)
I0607 05:14:47.155448 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 7.30622e-05 (* 0.0909091 = 6.64202e-06 loss)
I0607 05:14:47.155462 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 5.46961e-05 (* 0.0909091 = 4.97237e-06 loss)
I0607 05:14:47.155477 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 2.73752e-05 (* 0.0909091 = 2.48866e-06 loss)
I0607 05:14:47.155490 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 4.2839e-05 (* 0.0909091 = 3.89445e-06 loss)
I0607 05:14:47.155503 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.75
I0607 05:14:47.155525 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.75
I0607 05:14:47.155539 32403 solver.cpp:245] Train net output #149: total_confidence = 0.559125
I0607 05:14:47.155551 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.475602
I0607 05:14:47.155565 32403 sgd_solver.cpp:106] Iteration 19000, lr = 0.001
I0607 05:15:47.844130 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 47.2416 > 30) by scale factor 0.635033
I0607 05:16:45.111701 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.9201 > 30) by scale factor 0.939846
I0607 05:18:16.291780 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 48.2473 > 30) by scale factor 0.621797
I0607 05:18:30.202517 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.244 > 30) by scale factor 0.876067
I0607 05:18:34.070207 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 41.6418 > 30) by scale factor 0.720429
I0607 05:18:57.243753 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.5866 > 30) by scale factor 0.867388
I0607 05:20:16.824110 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 36.455 > 30) by scale factor 0.822932
I0607 05:21:11.666995 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 35.8118 > 30) by scale factor 0.837712
I0607 05:21:13.625962 32403 solver.cpp:229] Iteration 19500, loss = 3.90297
I0607 05:21:13.626044 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.414286
I0607 05:21:13.626063 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.75
I0607 05:21:13.626078 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.5
I0607 05:21:13.626092 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.625
I0607 05:21:13.626106 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.5
I0607 05:21:13.626118 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.25
I0607 05:21:13.626132 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.5
I0607 05:21:13.626145 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.625
I0607 05:21:13.626158 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.625
I0607 05:21:13.626171 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.75
I0607 05:21:13.626184 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.625
I0607 05:21:13.626199 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.75
I0607 05:21:13.626212 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.75
I0607 05:21:13.626225 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.625
I0607 05:21:13.626237 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.75
I0607 05:21:13.626250 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 0.75
I0607 05:21:13.626262 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 0.75
I0607 05:21:13.626274 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 0.875
I0607 05:21:13.626287 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 0.875
I0607 05:21:13.626299 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 0.875
I0607 05:21:13.626312 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 05:21:13.626324 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 05:21:13.626337 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 05:21:13.626348 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.75
I0607 05:21:13.626361 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.657143
I0607 05:21:13.626379 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.87388 (* 0.3 = 0.562163 loss)
I0607 05:21:13.626394 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.822944 (* 0.3 = 0.246883 loss)
I0607 05:21:13.626408 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.935061 (* 0.0272727 = 0.0255017 loss)
I0607 05:21:13.626423 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.5282 (* 0.0272727 = 0.0416781 loss)
I0607 05:21:13.626437 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.64341 (* 0.0272727 = 0.0448204 loss)
I0607 05:21:13.626451 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.55251 (* 0.0272727 = 0.0423413 loss)
I0607 05:21:13.626466 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 2.12775 (* 0.0272727 = 0.0580296 loss)
I0607 05:21:13.626479 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.58835 (* 0.0272727 = 0.0433187 loss)
I0607 05:21:13.626494 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 1.13987 (* 0.0272727 = 0.0310872 loss)
I0607 05:21:13.626509 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 1.14654 (* 0.0272727 = 0.0312693 loss)
I0607 05:21:13.626523 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 1.29633 (* 0.0272727 = 0.0353544 loss)
I0607 05:21:13.626538 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 1.02662 (* 0.0272727 = 0.0279987 loss)
I0607 05:21:13.626552 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.998852 (* 0.0272727 = 0.0272414 loss)
I0607 05:21:13.626566 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.977864 (* 0.0272727 = 0.026669 loss)
I0607 05:21:13.626621 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 1.07902 (* 0.0272727 = 0.0294279 loss)
I0607 05:21:13.626637 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.64499 (* 0.0272727 = 0.0175906 loss)
I0607 05:21:13.626652 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.548588 (* 0.0272727 = 0.0149615 loss)
I0607 05:21:13.626667 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.778408 (* 0.0272727 = 0.0212293 loss)
I0607 05:21:13.626680 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.50741 (* 0.0272727 = 0.0138384 loss)
I0607 05:21:13.626694 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.422484 (* 0.0272727 = 0.0115223 loss)
I0607 05:21:13.626709 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.477763 (* 0.0272727 = 0.0130299 loss)
I0607 05:21:13.626724 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.0221186 (* 0.0272727 = 0.000603236 loss)
I0607 05:21:13.626739 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.0158549 (* 0.0272727 = 0.000432407 loss)
I0607 05:21:13.626756 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.00629891 (* 0.0272727 = 0.000171788 loss)
I0607 05:21:13.626770 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.471429
I0607 05:21:13.626782 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.75
I0607 05:21:13.626796 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.75
I0607 05:21:13.626807 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.75
I0607 05:21:13.626819 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.625
I0607 05:21:13.626832 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.5
I0607 05:21:13.626843 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.625
I0607 05:21:13.626855 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.75
I0607 05:21:13.626868 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.625
I0607 05:21:13.626883 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.75
I0607 05:21:13.626896 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.625
I0607 05:21:13.626917 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.875
I0607 05:21:13.626929 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.75
I0607 05:21:13.626941 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.625
I0607 05:21:13.626953 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.75
I0607 05:21:13.626966 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 0.875
I0607 05:21:13.626987 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 0.75
I0607 05:21:13.626999 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 0.875
I0607 05:21:13.627012 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 0.875
I0607 05:21:13.627024 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 0.875
I0607 05:21:13.627037 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 05:21:13.627048 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 05:21:13.627060 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 05:21:13.627073 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.784091
I0607 05:21:13.627084 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.728571
I0607 05:21:13.627099 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.68184 (* 0.3 = 0.504553 loss)
I0607 05:21:13.627113 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.727576 (* 0.3 = 0.218273 loss)
I0607 05:21:13.627127 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.825175 (* 0.0272727 = 0.0225048 loss)
I0607 05:21:13.627152 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 1.00578 (* 0.0272727 = 0.0274305 loss)
I0607 05:21:13.627163 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 1.09855 (* 0.0272727 = 0.0299604 loss)
I0607 05:21:13.627173 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.22362 (* 0.0272727 = 0.0333715 loss)
I0607 05:21:13.627189 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.6708 (* 0.0272727 = 0.0455674 loss)
I0607 05:21:13.627207 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.57861 (* 0.0272727 = 0.0430529 loss)
I0607 05:21:13.627221 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 1.19633 (* 0.0272727 = 0.0326272 loss)
I0607 05:21:13.627235 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 1.20136 (* 0.0272727 = 0.0327643 loss)
I0607 05:21:13.627249 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 1.30155 (* 0.0272727 = 0.0354967 loss)
I0607 05:21:13.627264 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 1.13584 (* 0.0272727 = 0.0309774 loss)
I0607 05:21:13.627277 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 1.07435 (* 0.0272727 = 0.0293005 loss)
I0607 05:21:13.627291 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.701587 (* 0.0272727 = 0.0191342 loss)
I0607 05:21:13.627305 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 1.00502 (* 0.0272727 = 0.0274096 loss)
I0607 05:21:13.627320 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.565766 (* 0.0272727 = 0.01543 loss)
I0607 05:21:13.627334 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.354559 (* 0.0272727 = 0.0096698 loss)
I0607 05:21:13.627348 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.603611 (* 0.0272727 = 0.0164621 loss)
I0607 05:21:13.627362 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.418212 (* 0.0272727 = 0.0114058 loss)
I0607 05:21:13.627377 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.316188 (* 0.0272727 = 0.00862331 loss)
I0607 05:21:13.627391 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.653575 (* 0.0272727 = 0.0178248 loss)
I0607 05:21:13.627406 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.00584725 (* 0.0272727 = 0.00015947 loss)
I0607 05:21:13.627421 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.00315225 (* 0.0272727 = 8.59704e-05 loss)
I0607 05:21:13.627435 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 0.00342865 (* 0.0272727 = 9.35086e-05 loss)
I0607 05:21:13.627447 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.642857
I0607 05:21:13.627460 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 05:21:13.627472 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 0.75
I0607 05:21:13.627485 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 0.875
I0607 05:21:13.627496 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 0.875
I0607 05:21:13.627508 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 0.75
I0607 05:21:13.627521 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 1
I0607 05:21:13.627533 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.75
I0607 05:21:13.627545 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.875
I0607 05:21:13.627557 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.625
I0607 05:21:13.627569 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.625
I0607 05:21:13.627580 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.875
I0607 05:21:13.627593 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.625
I0607 05:21:13.627604 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 0.75
I0607 05:21:13.627616 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 0.75
I0607 05:21:13.627629 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 05:21:13.627650 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 0.75
I0607 05:21:13.627663 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 0.875
I0607 05:21:13.627676 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 0.875
I0607 05:21:13.627688 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 0.875
I0607 05:21:13.627701 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 05:21:13.627712 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 05:21:13.627724 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 05:21:13.627737 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.857955
I0607 05:21:13.627749 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.9
I0607 05:21:13.627763 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.9548 (* 1 = 0.9548 loss)
I0607 05:21:13.627777 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.427543 (* 1 = 0.427543 loss)
I0607 05:21:13.627792 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.414308 (* 0.0909091 = 0.0376644 loss)
I0607 05:21:13.627810 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.619043 (* 0.0909091 = 0.0562766 loss)
I0607 05:21:13.627825 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.608173 (* 0.0909091 = 0.0552885 loss)
I0607 05:21:13.627838 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.511702 (* 0.0909091 = 0.0465183 loss)
I0607 05:21:13.627852 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.689259 (* 0.0909091 = 0.0626599 loss)
I0607 05:21:13.627867 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.3827 (* 0.0909091 = 0.0347909 loss)
I0607 05:21:13.627882 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.982614 (* 0.0909091 = 0.0893285 loss)
I0607 05:21:13.627895 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.738589 (* 0.0909091 = 0.0671445 loss)
I0607 05:21:13.627909 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.732708 (* 0.0909091 = 0.0666098 loss)
I0607 05:21:13.627920 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 1.4237 (* 0.0909091 = 0.129428 loss)
I0607 05:21:13.627933 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.614763 (* 0.0909091 = 0.0558876 loss)
I0607 05:21:13.627948 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.789167 (* 0.0909091 = 0.0717425 loss)
I0607 05:21:13.627961 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.749966 (* 0.0909091 = 0.0681787 loss)
I0607 05:21:13.627976 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.518638 (* 0.0909091 = 0.0471489 loss)
I0607 05:21:13.627990 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.158668 (* 0.0909091 = 0.0144244 loss)
I0607 05:21:13.628005 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.587796 (* 0.0909091 = 0.053436 loss)
I0607 05:21:13.628018 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.45546 (* 0.0909091 = 0.0414055 loss)
I0607 05:21:13.628033 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.341069 (* 0.0909091 = 0.0310063 loss)
I0607 05:21:13.628047 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.368728 (* 0.0909091 = 0.0335207 loss)
I0607 05:21:13.628062 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.0187576 (* 0.0909091 = 0.00170524 loss)
I0607 05:21:13.628075 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.00338313 (* 0.0909091 = 0.000307557 loss)
I0607 05:21:13.628090 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 0.00215017 (* 0.0909091 = 0.00019547 loss)
I0607 05:21:13.628103 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.375
I0607 05:21:13.628114 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.375
I0607 05:21:13.628136 32403 solver.cpp:245] Train net output #149: total_confidence = 0.269343
I0607 05:21:13.628149 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.240297
I0607 05:21:13.628162 32403 sgd_solver.cpp:106] Iteration 19500, lr = 0.001
I0607 05:21:45.651391 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.0049 > 30) by scale factor 0.999837
I0607 05:22:02.629256 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.0504 > 30) by scale factor 0.809708
I0607 05:22:11.905359 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.0866 > 30) by scale factor 0.934969
I0607 05:22:52.816315 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.9238 > 30) by scale factor 0.79106
I0607 05:23:56.165355 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.7141 > 30) by scale factor 0.864201
I0607 05:24:41.766177 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.9194 > 30) by scale factor 0.911316
I0607 05:25:15.722941 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.589 > 30) by scale factor 0.949697
I0607 05:25:32.736491 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 46.7805 > 30) by scale factor 0.641292
I0607 05:26:10.674940 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.5859 > 30) by scale factor 0.980843
I0607 05:27:39.402600 32403 solver.cpp:456] Snapshotting to binary proto file /mnt/snapshots/mixed_lstm21_iter_20000.caffemodel
I0607 05:27:39.999873 32403 sgd_solver.cpp:273] Snapshotting solver state to binary proto file /mnt/snapshots/mixed_lstm21_iter_20000.solverstate
I0607 05:27:40.283576 32403 solver.cpp:338] Iteration 20000, Testing net (#0)
I0607 05:28:38.611060 32403 solver.cpp:393] Test loss: 2.56531
I0607 05:28:38.611217 32403 solver.cpp:406] Test net output #0: loss1/accuracy = 0.627389
I0607 05:28:38.611238 32403 solver.cpp:406] Test net output #1: loss1/accuracy01 = 0.797
I0607 05:28:38.611253 32403 solver.cpp:406] Test net output #2: loss1/accuracy02 = 0.666
I0607 05:28:38.611265 32403 solver.cpp:406] Test net output #3: loss1/accuracy03 = 0.566
I0607 05:28:38.611277 32403 solver.cpp:406] Test net output #4: loss1/accuracy04 = 0.486
I0607 05:28:38.611290 32403 solver.cpp:406] Test net output #5: loss1/accuracy05 = 0.542
I0607 05:28:38.611302 32403 solver.cpp:406] Test net output #6: loss1/accuracy06 = 0.716
I0607 05:28:38.611315 32403 solver.cpp:406] Test net output #7: loss1/accuracy07 = 0.849
I0607 05:28:38.611327 32403 solver.cpp:406] Test net output #8: loss1/accuracy08 = 0.921
I0607 05:28:38.611341 32403 solver.cpp:406] Test net output #9: loss1/accuracy09 = 0.966
I0607 05:28:38.611353 32403 solver.cpp:406] Test net output #10: loss1/accuracy10 = 0.985
I0607 05:28:38.611366 32403 solver.cpp:406] Test net output #11: loss1/accuracy11 = 0.996
I0607 05:28:38.611379 32403 solver.cpp:406] Test net output #12: loss1/accuracy12 = 0.999
I0607 05:28:38.611392 32403 solver.cpp:406] Test net output #13: loss1/accuracy13 = 1
I0607 05:28:38.611404 32403 solver.cpp:406] Test net output #14: loss1/accuracy14 = 1
I0607 05:28:38.611426 32403 solver.cpp:406] Test net output #15: loss1/accuracy15 = 1
I0607 05:28:38.611446 32403 solver.cpp:406] Test net output #16: loss1/accuracy16 = 1
I0607 05:28:38.611460 32403 solver.cpp:406] Test net output #17: loss1/accuracy17 = 1
I0607 05:28:38.611472 32403 solver.cpp:406] Test net output #18: loss1/accuracy18 = 1
I0607 05:28:38.611485 32403 solver.cpp:406] Test net output #19: loss1/accuracy19 = 1
I0607 05:28:38.611495 32403 solver.cpp:406] Test net output #20: loss1/accuracy20 = 1
I0607 05:28:38.611507 32403 solver.cpp:406] Test net output #21: loss1/accuracy21 = 1
I0607 05:28:38.611520 32403 solver.cpp:406] Test net output #22: loss1/accuracy22 = 1
I0607 05:28:38.611531 32403 solver.cpp:406] Test net output #23: loss1/accuracy_incl_empty = 0.892911
I0607 05:28:38.611542 32403 solver.cpp:406] Test net output #24: loss1/accuracy_top3 = 0.848355
I0607 05:28:38.611559 32403 solver.cpp:406] Test net output #25: loss1/cross_entropy_loss = 1.37807 (* 0.3 = 0.413422 loss)
I0607 05:28:38.611575 32403 solver.cpp:406] Test net output #26: loss1/cross_entropy_loss_incl_empty = 0.394972 (* 0.3 = 0.118492 loss)
I0607 05:28:38.611590 32403 solver.cpp:406] Test net output #27: loss1/loss01 = 0.870187 (* 0.0272727 = 0.0237324 loss)
I0607 05:28:38.611604 32403 solver.cpp:406] Test net output #28: loss1/loss02 = 1.28792 (* 0.0272727 = 0.0351251 loss)
I0607 05:28:38.611618 32403 solver.cpp:406] Test net output #29: loss1/loss03 = 1.6013 (* 0.0272727 = 0.0436717 loss)
I0607 05:28:38.611631 32403 solver.cpp:406] Test net output #30: loss1/loss04 = 1.78077 (* 0.0272727 = 0.0485665 loss)
I0607 05:28:38.611645 32403 solver.cpp:406] Test net output #31: loss1/loss05 = 1.52383 (* 0.0272727 = 0.0415589 loss)
I0607 05:28:38.611660 32403 solver.cpp:406] Test net output #32: loss1/loss06 = 0.971692 (* 0.0272727 = 0.0265007 loss)
I0607 05:28:38.611672 32403 solver.cpp:406] Test net output #33: loss1/loss07 = 0.530513 (* 0.0272727 = 0.0144685 loss)
I0607 05:28:38.611686 32403 solver.cpp:406] Test net output #34: loss1/loss08 = 0.301789 (* 0.0272727 = 0.0082306 loss)
I0607 05:28:38.611701 32403 solver.cpp:406] Test net output #35: loss1/loss09 = 0.172021 (* 0.0272727 = 0.00469148 loss)
I0607 05:28:38.611716 32403 solver.cpp:406] Test net output #36: loss1/loss10 = 0.0738178 (* 0.0272727 = 0.00201321 loss)
I0607 05:28:38.611729 32403 solver.cpp:406] Test net output #37: loss1/loss11 = 0.0205861 (* 0.0272727 = 0.000561439 loss)
I0607 05:28:38.611743 32403 solver.cpp:406] Test net output #38: loss1/loss12 = 0.0113544 (* 0.0272727 = 0.000309665 loss)
I0607 05:28:38.611757 32403 solver.cpp:406] Test net output #39: loss1/loss13 = 0.00729304 (* 0.0272727 = 0.000198901 loss)
I0607 05:28:38.611789 32403 solver.cpp:406] Test net output #40: loss1/loss14 = 0.0051926 (* 0.0272727 = 0.000141616 loss)
I0607 05:28:38.611806 32403 solver.cpp:406] Test net output #41: loss1/loss15 = 0.00348129 (* 0.0272727 = 9.49443e-05 loss)
I0607 05:28:38.611821 32403 solver.cpp:406] Test net output #42: loss1/loss16 = 0.00264375 (* 0.0272727 = 7.21024e-05 loss)
I0607 05:28:38.611835 32403 solver.cpp:406] Test net output #43: loss1/loss17 = 0.00231906 (* 0.0272727 = 6.32471e-05 loss)
I0607 05:28:38.611850 32403 solver.cpp:406] Test net output #44: loss1/loss18 = 0.00195568 (* 0.0272727 = 5.33369e-05 loss)
I0607 05:28:38.611863 32403 solver.cpp:406] Test net output #45: loss1/loss19 = 0.00148808 (* 0.0272727 = 4.0584e-05 loss)
I0607 05:28:38.611879 32403 solver.cpp:406] Test net output #46: loss1/loss20 = 0.00133036 (* 0.0272727 = 3.62826e-05 loss)
I0607 05:28:38.611894 32403 solver.cpp:406] Test net output #47: loss1/loss21 = 0.00098117 (* 0.0272727 = 2.67592e-05 loss)
I0607 05:28:38.611908 32403 solver.cpp:406] Test net output #48: loss1/loss22 = 0.000963362 (* 0.0272727 = 2.62735e-05 loss)
I0607 05:28:38.611920 32403 solver.cpp:406] Test net output #49: loss2/accuracy = 0.779149
I0607 05:28:38.611932 32403 solver.cpp:406] Test net output #50: loss2/accuracy01 = 0.888
I0607 05:28:38.611944 32403 solver.cpp:406] Test net output #51: loss2/accuracy02 = 0.837
I0607 05:28:38.611955 32403 solver.cpp:406] Test net output #52: loss2/accuracy03 = 0.767
I0607 05:28:38.611968 32403 solver.cpp:406] Test net output #53: loss2/accuracy04 = 0.661
I0607 05:28:38.611979 32403 solver.cpp:406] Test net output #54: loss2/accuracy05 = 0.664
I0607 05:28:38.611990 32403 solver.cpp:406] Test net output #55: loss2/accuracy06 = 0.771
I0607 05:28:38.612002 32403 solver.cpp:406] Test net output #56: loss2/accuracy07 = 0.875
I0607 05:28:38.612013 32403 solver.cpp:406] Test net output #57: loss2/accuracy08 = 0.934
I0607 05:28:38.612025 32403 solver.cpp:406] Test net output #58: loss2/accuracy09 = 0.964
I0607 05:28:38.612036 32403 solver.cpp:406] Test net output #59: loss2/accuracy10 = 0.983
I0607 05:28:38.612047 32403 solver.cpp:406] Test net output #60: loss2/accuracy11 = 0.995
I0607 05:28:38.612058 32403 solver.cpp:406] Test net output #61: loss2/accuracy12 = 0.999
I0607 05:28:38.612071 32403 solver.cpp:406] Test net output #62: loss2/accuracy13 = 0.999
I0607 05:28:38.612082 32403 solver.cpp:406] Test net output #63: loss2/accuracy14 = 1
I0607 05:28:38.612092 32403 solver.cpp:406] Test net output #64: loss2/accuracy15 = 1
I0607 05:28:38.612103 32403 solver.cpp:406] Test net output #65: loss2/accuracy16 = 1
I0607 05:28:38.612114 32403 solver.cpp:406] Test net output #66: loss2/accuracy17 = 1
I0607 05:28:38.612125 32403 solver.cpp:406] Test net output #67: loss2/accuracy18 = 1
I0607 05:28:38.612136 32403 solver.cpp:406] Test net output #68: loss2/accuracy19 = 1
I0607 05:28:38.612149 32403 solver.cpp:406] Test net output #69: loss2/accuracy20 = 1
I0607 05:28:38.612159 32403 solver.cpp:406] Test net output #70: loss2/accuracy21 = 1
I0607 05:28:38.612170 32403 solver.cpp:406] Test net output #71: loss2/accuracy22 = 1
I0607 05:28:38.612181 32403 solver.cpp:406] Test net output #72: loss2/accuracy_incl_empty = 0.933682
I0607 05:28:38.612192 32403 solver.cpp:406] Test net output #73: loss2/accuracy_top3 = 0.911095
I0607 05:28:38.612206 32403 solver.cpp:406] Test net output #74: loss2/cross_entropy_loss = 0.904612 (* 0.3 = 0.271384 loss)
I0607 05:28:38.612221 32403 solver.cpp:406] Test net output #75: loss2/cross_entropy_loss_incl_empty = 0.267433 (* 0.3 = 0.0802298 loss)
I0607 05:28:38.612237 32403 solver.cpp:406] Test net output #76: loss2/loss01 = 0.575113 (* 0.0272727 = 0.0156849 loss)
I0607 05:28:38.612251 32403 solver.cpp:406] Test net output #77: loss2/loss02 = 0.778748 (* 0.0272727 = 0.0212386 loss)
I0607 05:28:38.612277 32403 solver.cpp:406] Test net output #78: loss2/loss03 = 0.950921 (* 0.0272727 = 0.0259342 loss)
I0607 05:28:38.612293 32403 solver.cpp:406] Test net output #79: loss2/loss04 = 1.19645 (* 0.0272727 = 0.0326304 loss)
I0607 05:28:38.612306 32403 solver.cpp:406] Test net output #80: loss2/loss05 = 1.09846 (* 0.0272727 = 0.0299579 loss)
I0607 05:28:38.612319 32403 solver.cpp:406] Test net output #81: loss2/loss06 = 0.75062 (* 0.0272727 = 0.0204715 loss)
I0607 05:28:38.612334 32403 solver.cpp:406] Test net output #82: loss2/loss07 = 0.423901 (* 0.0272727 = 0.0115609 loss)
I0607 05:28:38.612347 32403 solver.cpp:406] Test net output #83: loss2/loss08 = 0.234881 (* 0.0272727 = 0.00640585 loss)
I0607 05:28:38.612361 32403 solver.cpp:406] Test net output #84: loss2/loss09 = 0.14216 (* 0.0272727 = 0.00387708 loss)
I0607 05:28:38.612375 32403 solver.cpp:406] Test net output #85: loss2/loss10 = 0.0675224 (* 0.0272727 = 0.00184152 loss)
I0607 05:28:38.612388 32403 solver.cpp:406] Test net output #86: loss2/loss11 = 0.0184056 (* 0.0272727 = 0.000501972 loss)
I0607 05:28:38.612402 32403 solver.cpp:406] Test net output #87: loss2/loss12 = 0.00811854 (* 0.0272727 = 0.000221415 loss)
I0607 05:28:38.612416 32403 solver.cpp:406] Test net output #88: loss2/loss13 = 0.00471783 (* 0.0272727 = 0.000128668 loss)
I0607 05:28:38.612431 32403 solver.cpp:406] Test net output #89: loss2/loss14 = 0.00282666 (* 0.0272727 = 7.70906e-05 loss)
I0607 05:28:38.612443 32403 solver.cpp:406] Test net output #90: loss2/loss15 = 0.00181173 (* 0.0272727 = 4.94109e-05 loss)
I0607 05:28:38.612457 32403 solver.cpp:406] Test net output #91: loss2/loss16 = 0.00129049 (* 0.0272727 = 3.51953e-05 loss)
I0607 05:28:38.612471 32403 solver.cpp:406] Test net output #92: loss2/loss17 = 0.00107107 (* 0.0272727 = 2.9211e-05 loss)
I0607 05:28:38.612485 32403 solver.cpp:406] Test net output #93: loss2/loss18 = 0.000932851 (* 0.0272727 = 2.54414e-05 loss)
I0607 05:28:38.612499 32403 solver.cpp:406] Test net output #94: loss2/loss19 = 0.000662467 (* 0.0272727 = 1.80673e-05 loss)
I0607 05:28:38.612512 32403 solver.cpp:406] Test net output #95: loss2/loss20 = 0.000741866 (* 0.0272727 = 2.02327e-05 loss)
I0607 05:28:38.612526 32403 solver.cpp:406] Test net output #96: loss2/loss21 = 0.000684425 (* 0.0272727 = 1.86661e-05 loss)
I0607 05:28:38.612540 32403 solver.cpp:406] Test net output #97: loss2/loss22 = 0.000594999 (* 0.0272727 = 1.62272e-05 loss)
I0607 05:28:38.612552 32403 solver.cpp:406] Test net output #98: loss3/accuracy = 0.857905
I0607 05:28:38.612563 32403 solver.cpp:406] Test net output #99: loss3/accuracy01 = 0.901
I0607 05:28:38.612576 32403 solver.cpp:406] Test net output #100: loss3/accuracy02 = 0.864
I0607 05:28:38.612586 32403 solver.cpp:406] Test net output #101: loss3/accuracy03 = 0.859
I0607 05:28:38.612598 32403 solver.cpp:406] Test net output #102: loss3/accuracy04 = 0.848
I0607 05:28:38.612609 32403 solver.cpp:406] Test net output #103: loss3/accuracy05 = 0.841
I0607 05:28:38.612622 32403 solver.cpp:406] Test net output #104: loss3/accuracy06 = 0.872
I0607 05:28:38.612632 32403 solver.cpp:406] Test net output #105: loss3/accuracy07 = 0.906
I0607 05:28:38.612643 32403 solver.cpp:406] Test net output #106: loss3/accuracy08 = 0.949
I0607 05:28:38.612655 32403 solver.cpp:406] Test net output #107: loss3/accuracy09 = 0.97
I0607 05:28:38.612666 32403 solver.cpp:406] Test net output #108: loss3/accuracy10 = 0.984
I0607 05:28:38.612679 32403 solver.cpp:406] Test net output #109: loss3/accuracy11 = 0.994
I0607 05:28:38.612689 32403 solver.cpp:406] Test net output #110: loss3/accuracy12 = 0.997
I0607 05:28:38.612700 32403 solver.cpp:406] Test net output #111: loss3/accuracy13 = 0.999
I0607 05:28:38.612709 32403 solver.cpp:406] Test net output #112: loss3/accuracy14 = 0.999
I0607 05:28:38.612715 32403 solver.cpp:406] Test net output #113: loss3/accuracy15 = 1
I0607 05:28:38.612722 32403 solver.cpp:406] Test net output #114: loss3/accuracy16 = 1
I0607 05:28:38.612745 32403 solver.cpp:406] Test net output #115: loss3/accuracy17 = 1
I0607 05:28:38.612757 32403 solver.cpp:406] Test net output #116: loss3/accuracy18 = 1
I0607 05:28:38.612768 32403 solver.cpp:406] Test net output #117: loss3/accuracy19 = 1
I0607 05:28:38.612779 32403 solver.cpp:406] Test net output #118: loss3/accuracy20 = 1
I0607 05:28:38.612790 32403 solver.cpp:406] Test net output #119: loss3/accuracy21 = 1
I0607 05:28:38.612802 32403 solver.cpp:406] Test net output #120: loss3/accuracy22 = 1
I0607 05:28:38.612813 32403 solver.cpp:406] Test net output #121: loss3/accuracy_incl_empty = 0.954364
I0607 05:28:38.612824 32403 solver.cpp:406] Test net output #122: loss3/accuracy_top3 = 0.930666
I0607 05:28:38.612838 32403 solver.cpp:406] Test net output #123: loss3/cross_entropy_loss = 0.64598 (* 1 = 0.64598 loss)
I0607 05:28:38.612851 32403 solver.cpp:406] Test net output #124: loss3/cross_entropy_loss_incl_empty = 0.210458 (* 1 = 0.210458 loss)
I0607 05:28:38.612865 32403 solver.cpp:406] Test net output #125: loss3/loss01 = 0.483595 (* 0.0909091 = 0.0439632 loss)
I0607 05:28:38.612879 32403 solver.cpp:406] Test net output #126: loss3/loss02 = 0.597804 (* 0.0909091 = 0.0543458 loss)
I0607 05:28:38.612893 32403 solver.cpp:406] Test net output #127: loss3/loss03 = 0.613951 (* 0.0909091 = 0.0558137 loss)
I0607 05:28:38.612907 32403 solver.cpp:406] Test net output #128: loss3/loss04 = 0.699622 (* 0.0909091 = 0.063602 loss)
I0607 05:28:38.612920 32403 solver.cpp:406] Test net output #129: loss3/loss05 = 0.689037 (* 0.0909091 = 0.0626397 loss)
I0607 05:28:38.612937 32403 solver.cpp:406] Test net output #130: loss3/loss06 = 0.541247 (* 0.0909091 = 0.0492043 loss)
I0607 05:28:38.612951 32403 solver.cpp:406] Test net output #131: loss3/loss07 = 0.36446 (* 0.0909091 = 0.0331327 loss)
I0607 05:28:38.612965 32403 solver.cpp:406] Test net output #132: loss3/loss08 = 0.205465 (* 0.0909091 = 0.0186787 loss)
I0607 05:28:38.612978 32403 solver.cpp:406] Test net output #133: loss3/loss09 = 0.131117 (* 0.0909091 = 0.0119197 loss)
I0607 05:28:38.612993 32403 solver.cpp:406] Test net output #134: loss3/loss10 = 0.0720097 (* 0.0909091 = 0.00654634 loss)
I0607 05:28:38.613005 32403 solver.cpp:406] Test net output #135: loss3/loss11 = 0.0238907 (* 0.0909091 = 0.00217188 loss)
I0607 05:28:38.613019 32403 solver.cpp:406] Test net output #136: loss3/loss12 = 0.0118998 (* 0.0909091 = 0.0010818 loss)
I0607 05:28:38.613034 32403 solver.cpp:406] Test net output #137: loss3/loss13 = 0.00635881 (* 0.0909091 = 0.000578074 loss)
I0607 05:28:38.613047 32403 solver.cpp:406] Test net output #138: loss3/loss14 = 0.00361755 (* 0.0909091 = 0.000328868 loss)
I0607 05:28:38.613060 32403 solver.cpp:406] Test net output #139: loss3/loss15 = 0.00135419 (* 0.0909091 = 0.000123108 loss)
I0607 05:28:38.613075 32403 solver.cpp:406] Test net output #140: loss3/loss16 = 0.000889974 (* 0.0909091 = 8.09067e-05 loss)
I0607 05:28:38.613088 32403 solver.cpp:406] Test net output #141: loss3/loss17 = 0.000531326 (* 0.0909091 = 4.83024e-05 loss)
I0607 05:28:38.613101 32403 solver.cpp:406] Test net output #142: loss3/loss18 = 0.000412361 (* 0.0909091 = 3.74873e-05 loss)
I0607 05:28:38.613116 32403 solver.cpp:406] Test net output #143: loss3/loss19 = 0.000324758 (* 0.0909091 = 2.95235e-05 loss)
I0607 05:28:38.613144 32403 solver.cpp:406] Test net output #144: loss3/loss20 = 0.000338155 (* 0.0909091 = 3.07414e-05 loss)
I0607 05:28:38.613159 32403 solver.cpp:406] Test net output #145: loss3/loss21 = 0.000320841 (* 0.0909091 = 2.91673e-05 loss)
I0607 05:28:38.613173 32403 solver.cpp:406] Test net output #146: loss3/loss22 = 0.000309537 (* 0.0909091 = 2.81397e-05 loss)
I0607 05:28:38.613184 32403 solver.cpp:406] Test net output #147: total_accuracy = 0.617
I0607 05:28:38.613195 32403 solver.cpp:406] Test net output #148: total_accuracy_not_rec = 0.584
I0607 05:28:38.613207 32403 solver.cpp:406] Test net output #149: total_confidence = 0.53301
I0607 05:28:38.613231 32403 solver.cpp:406] Test net output #150: total_confidence_not_rec = 0.505856
I0607 05:28:38.613245 32403 solver.cpp:338] Iteration 20000, Testing net (#1)
I0607 05:29:37.046875 32403 solver.cpp:393] Test loss: 3.66497
I0607 05:29:37.047039 32403 solver.cpp:406] Test net output #0: loss1/accuracy = 0.57156
I0607 05:29:37.047070 32403 solver.cpp:406] Test net output #1: loss1/accuracy01 = 0.773
I0607 05:29:37.047092 32403 solver.cpp:406] Test net output #2: loss1/accuracy02 = 0.651
I0607 05:29:37.047112 32403 solver.cpp:406] Test net output #3: loss1/accuracy03 = 0.53
I0607 05:29:37.047132 32403 solver.cpp:406] Test net output #4: loss1/accuracy04 = 0.392
I0607 05:29:37.047153 32403 solver.cpp:406] Test net output #5: loss1/accuracy05 = 0.487
I0607 05:29:37.047173 32403 solver.cpp:406] Test net output #6: loss1/accuracy06 = 0.644
I0607 05:29:37.047194 32403 solver.cpp:406] Test net output #7: loss1/accuracy07 = 0.743
I0607 05:29:37.047214 32403 solver.cpp:406] Test net output #8: loss1/accuracy08 = 0.801
I0607 05:29:37.047232 32403 solver.cpp:406] Test net output #9: loss1/accuracy09 = 0.84
I0607 05:29:37.047253 32403 solver.cpp:406] Test net output #10: loss1/accuracy10 = 0.851
I0607 05:29:37.047276 32403 solver.cpp:406] Test net output #11: loss1/accuracy11 = 0.889
I0607 05:29:37.047296 32403 solver.cpp:406] Test net output #12: loss1/accuracy12 = 0.9
I0607 05:29:37.047314 32403 solver.cpp:406] Test net output #13: loss1/accuracy13 = 0.928
I0607 05:29:37.047336 32403 solver.cpp:406] Test net output #14: loss1/accuracy14 = 0.937
I0607 05:29:37.047355 32403 solver.cpp:406] Test net output #15: loss1/accuracy15 = 0.953
I0607 05:29:37.047374 32403 solver.cpp:406] Test net output #16: loss1/accuracy16 = 0.966
I0607 05:29:37.047394 32403 solver.cpp:406] Test net output #17: loss1/accuracy17 = 0.982
I0607 05:29:37.047413 32403 solver.cpp:406] Test net output #18: loss1/accuracy18 = 0.986
I0607 05:29:37.047433 32403 solver.cpp:406] Test net output #19: loss1/accuracy19 = 0.988
I0607 05:29:37.047451 32403 solver.cpp:406] Test net output #20: loss1/accuracy20 = 0.995
I0607 05:29:37.047472 32403 solver.cpp:406] Test net output #21: loss1/accuracy21 = 0.998
I0607 05:29:37.047489 32403 solver.cpp:406] Test net output #22: loss1/accuracy22 = 1
I0607 05:29:37.047509 32403 solver.cpp:406] Test net output #23: loss1/accuracy_incl_empty = 0.840092
I0607 05:29:37.047528 32403 solver.cpp:406] Test net output #24: loss1/accuracy_top3 = 0.79421
I0607 05:29:37.047554 32403 solver.cpp:406] Test net output #25: loss1/cross_entropy_loss = 1.6131 (* 0.3 = 0.48393 loss)
I0607 05:29:37.047579 32403 solver.cpp:406] Test net output #26: loss1/cross_entropy_loss_incl_empty = 0.593302 (* 0.3 = 0.177991 loss)
I0607 05:29:37.047605 32403 solver.cpp:406] Test net output #27: loss1/loss01 = 0.977307 (* 0.0272727 = 0.0266538 loss)
I0607 05:29:37.047631 32403 solver.cpp:406] Test net output #28: loss1/loss02 = 1.37846 (* 0.0272727 = 0.0375943 loss)
I0607 05:29:37.047654 32403 solver.cpp:406] Test net output #29: loss1/loss03 = 1.69571 (* 0.0272727 = 0.0462465 loss)
I0607 05:29:37.047677 32403 solver.cpp:406] Test net output #30: loss1/loss04 = 2.00327 (* 0.0272727 = 0.0546345 loss)
I0607 05:29:37.047699 32403 solver.cpp:406] Test net output #31: loss1/loss05 = 1.69864 (* 0.0272727 = 0.0463266 loss)
I0607 05:29:37.047722 32403 solver.cpp:406] Test net output #32: loss1/loss06 = 1.29018 (* 0.0272727 = 0.0351867 loss)
I0607 05:29:37.047746 32403 solver.cpp:406] Test net output #33: loss1/loss07 = 0.877677 (* 0.0272727 = 0.0239367 loss)
I0607 05:29:37.047770 32403 solver.cpp:406] Test net output #34: loss1/loss08 = 0.727236 (* 0.0272727 = 0.0198337 loss)
I0607 05:29:37.047791 32403 solver.cpp:406] Test net output #35: loss1/loss09 = 0.607739 (* 0.0272727 = 0.0165747 loss)
I0607 05:29:37.047823 32403 solver.cpp:406] Test net output #36: loss1/loss10 = 0.544626 (* 0.0272727 = 0.0148534 loss)
I0607 05:29:37.047847 32403 solver.cpp:406] Test net output #37: loss1/loss11 = 0.463321 (* 0.0272727 = 0.012636 loss)
I0607 05:29:37.047880 32403 solver.cpp:406] Test net output #38: loss1/loss12 = 0.391454 (* 0.0272727 = 0.010676 loss)
I0607 05:29:37.047905 32403 solver.cpp:406] Test net output #39: loss1/loss13 = 0.26475 (* 0.0272727 = 0.00722044 loss)
I0607 05:29:37.047955 32403 solver.cpp:406] Test net output #40: loss1/loss14 = 0.230797 (* 0.0272727 = 0.00629446 loss)
I0607 05:29:37.047981 32403 solver.cpp:406] Test net output #41: loss1/loss15 = 0.176346 (* 0.0272727 = 0.00480945 loss)
I0607 05:29:37.048005 32403 solver.cpp:406] Test net output #42: loss1/loss16 = 0.149103 (* 0.0272727 = 0.00406646 loss)
I0607 05:29:37.048029 32403 solver.cpp:406] Test net output #43: loss1/loss17 = 0.0840473 (* 0.0272727 = 0.0022922 loss)
I0607 05:29:37.048058 32403 solver.cpp:406] Test net output #44: loss1/loss18 = 0.0719268 (* 0.0272727 = 0.00196164 loss)
I0607 05:29:37.048082 32403 solver.cpp:406] Test net output #45: loss1/loss19 = 0.0638056 (* 0.0272727 = 0.00174015 loss)
I0607 05:29:37.048106 32403 solver.cpp:406] Test net output #46: loss1/loss20 = 0.0311972 (* 0.0272727 = 0.000850832 loss)
I0607 05:29:37.048130 32403 solver.cpp:406] Test net output #47: loss1/loss21 = 0.0155019 (* 0.0272727 = 0.00042278 loss)
I0607 05:29:37.048152 32403 solver.cpp:406] Test net output #48: loss1/loss22 = 0.00189825 (* 0.0272727 = 5.17704e-05 loss)
I0607 05:29:37.048173 32403 solver.cpp:406] Test net output #49: loss2/accuracy = 0.697198
I0607 05:29:37.048194 32403 solver.cpp:406] Test net output #50: loss2/accuracy01 = 0.852
I0607 05:29:37.048221 32403 solver.cpp:406] Test net output #51: loss2/accuracy02 = 0.793
I0607 05:29:37.048240 32403 solver.cpp:406] Test net output #52: loss2/accuracy03 = 0.729
I0607 05:29:37.048259 32403 solver.cpp:406] Test net output #53: loss2/accuracy04 = 0.604
I0607 05:29:37.048282 32403 solver.cpp:406] Test net output #54: loss2/accuracy05 = 0.633
I0607 05:29:37.048301 32403 solver.cpp:406] Test net output #55: loss2/accuracy06 = 0.725
I0607 05:29:37.048321 32403 solver.cpp:406] Test net output #56: loss2/accuracy07 = 0.77
I0607 05:29:37.048341 32403 solver.cpp:406] Test net output #57: loss2/accuracy08 = 0.811
I0607 05:29:37.048362 32403 solver.cpp:406] Test net output #58: loss2/accuracy09 = 0.855
I0607 05:29:37.048383 32403 solver.cpp:406] Test net output #59: loss2/accuracy10 = 0.868
I0607 05:29:37.048403 32403 solver.cpp:406] Test net output #60: loss2/accuracy11 = 0.9
I0607 05:29:37.048424 32403 solver.cpp:406] Test net output #61: loss2/accuracy12 = 0.908
I0607 05:29:37.048444 32403 solver.cpp:406] Test net output #62: loss2/accuracy13 = 0.933
I0607 05:29:37.048462 32403 solver.cpp:406] Test net output #63: loss2/accuracy14 = 0.94
I0607 05:29:37.048481 32403 solver.cpp:406] Test net output #64: loss2/accuracy15 = 0.951
I0607 05:29:37.048501 32403 solver.cpp:406] Test net output #65: loss2/accuracy16 = 0.967
I0607 05:29:37.048521 32403 solver.cpp:406] Test net output #66: loss2/accuracy17 = 0.982
I0607 05:29:37.048542 32403 solver.cpp:406] Test net output #67: loss2/accuracy18 = 0.986
I0607 05:29:37.048560 32403 solver.cpp:406] Test net output #68: loss2/accuracy19 = 0.988
I0607 05:29:37.048580 32403 solver.cpp:406] Test net output #69: loss2/accuracy20 = 0.995
I0607 05:29:37.048600 32403 solver.cpp:406] Test net output #70: loss2/accuracy21 = 0.998
I0607 05:29:37.048620 32403 solver.cpp:406] Test net output #71: loss2/accuracy22 = 1
I0607 05:29:37.048640 32403 solver.cpp:406] Test net output #72: loss2/accuracy_incl_empty = 0.879364
I0607 05:29:37.048660 32403 solver.cpp:406] Test net output #73: loss2/accuracy_top3 = 0.866708
I0607 05:29:37.048683 32403 solver.cpp:406] Test net output #74: loss2/cross_entropy_loss = 1.18194 (* 0.3 = 0.354581 loss)
I0607 05:29:37.048707 32403 solver.cpp:406] Test net output #75: loss2/cross_entropy_loss_incl_empty = 0.455585 (* 0.3 = 0.136675 loss)
I0607 05:29:37.048732 32403 solver.cpp:406] Test net output #76: loss2/loss01 = 0.725399 (* 0.0272727 = 0.0197836 loss)
I0607 05:29:37.048754 32403 solver.cpp:406] Test net output #77: loss2/loss02 = 0.957471 (* 0.0272727 = 0.0261129 loss)
I0607 05:29:37.048795 32403 solver.cpp:406] Test net output #78: loss2/loss03 = 1.08129 (* 0.0272727 = 0.0294897 loss)
I0607 05:29:37.048821 32403 solver.cpp:406] Test net output #79: loss2/loss04 = 1.36558 (* 0.0272727 = 0.0372431 loss)
I0607 05:29:37.048845 32403 solver.cpp:406] Test net output #80: loss2/loss05 = 1.31323 (* 0.0272727 = 0.0358155 loss)
I0607 05:29:37.048868 32403 solver.cpp:406] Test net output #81: loss2/loss06 = 1.01173 (* 0.0272727 = 0.0275925 loss)
I0607 05:29:37.048892 32403 solver.cpp:406] Test net output #82: loss2/loss07 = 0.77666 (* 0.0272727 = 0.0211816 loss)
I0607 05:29:37.048914 32403 solver.cpp:406] Test net output #83: loss2/loss08 = 0.630099 (* 0.0272727 = 0.0171845 loss)
I0607 05:29:37.048943 32403 solver.cpp:406] Test net output #84: loss2/loss09 = 0.540302 (* 0.0272727 = 0.0147355 loss)
I0607 05:29:37.048966 32403 solver.cpp:406] Test net output #85: loss2/loss10 = 0.4767 (* 0.0272727 = 0.0130009 loss)
I0607 05:29:37.048990 32403 solver.cpp:406] Test net output #86: loss2/loss11 = 0.408363 (* 0.0272727 = 0.0111372 loss)
I0607 05:29:37.049015 32403 solver.cpp:406] Test net output #87: loss2/loss12 = 0.332928 (* 0.0272727 = 0.00907986 loss)
I0607 05:29:37.049038 32403 solver.cpp:406] Test net output #88: loss2/loss13 = 0.234569 (* 0.0272727 = 0.00639735 loss)
I0607 05:29:37.049062 32403 solver.cpp:406] Test net output #89: loss2/loss14 = 0.204193 (* 0.0272727 = 0.0055689 loss)
I0607 05:29:37.049087 32403 solver.cpp:406] Test net output #90: loss2/loss15 = 0.161853 (* 0.0272727 = 0.00441417 loss)
I0607 05:29:37.049132 32403 solver.cpp:406] Test net output #91: loss2/loss16 = 0.139305 (* 0.0272727 = 0.00379922 loss)
I0607 05:29:37.049159 32403 solver.cpp:406] Test net output #92: loss2/loss17 = 0.0797339 (* 0.0272727 = 0.00217456 loss)
I0607 05:29:37.049185 32403 solver.cpp:406] Test net output #93: loss2/loss18 = 0.0619904 (* 0.0272727 = 0.00169065 loss)
I0607 05:29:37.049208 32403 solver.cpp:406] Test net output #94: loss2/loss19 = 0.0644867 (* 0.0272727 = 0.00175873 loss)
I0607 05:29:37.049232 32403 solver.cpp:406] Test net output #95: loss2/loss20 = 0.0262901 (* 0.0272727 = 0.000717003 loss)
I0607 05:29:37.049257 32403 solver.cpp:406] Test net output #96: loss2/loss21 = 0.0135738 (* 0.0272727 = 0.000370194 loss)
I0607 05:29:37.049281 32403 solver.cpp:406] Test net output #97: loss2/loss22 = 0.00158695 (* 0.0272727 = 4.32805e-05 loss)
I0607 05:29:37.049304 32403 solver.cpp:406] Test net output #98: loss3/accuracy = 0.813159
I0607 05:29:37.049324 32403 solver.cpp:406] Test net output #99: loss3/accuracy01 = 0.876
I0607 05:29:37.049345 32403 solver.cpp:406] Test net output #100: loss3/accuracy02 = 0.852
I0607 05:29:37.049365 32403 solver.cpp:406] Test net output #101: loss3/accuracy03 = 0.85
I0607 05:29:37.049386 32403 solver.cpp:406] Test net output #102: loss3/accuracy04 = 0.827
I0607 05:29:37.049404 32403 solver.cpp:406] Test net output #103: loss3/accuracy05 = 0.827
I0607 05:29:37.049424 32403 solver.cpp:406] Test net output #104: loss3/accuracy06 = 0.834
I0607 05:29:37.049444 32403 solver.cpp:406] Test net output #105: loss3/accuracy07 = 0.868
I0607 05:29:37.049465 32403 solver.cpp:406] Test net output #106: loss3/accuracy08 = 0.891
I0607 05:29:37.049481 32403 solver.cpp:406] Test net output #107: loss3/accuracy09 = 0.89
I0607 05:29:37.049504 32403 solver.cpp:406] Test net output #108: loss3/accuracy10 = 0.898
I0607 05:29:37.049523 32403 solver.cpp:406] Test net output #109: loss3/accuracy11 = 0.935
I0607 05:29:37.049545 32403 solver.cpp:406] Test net output #110: loss3/accuracy12 = 0.925
I0607 05:29:37.049566 32403 solver.cpp:406] Test net output #111: loss3/accuracy13 = 0.945
I0607 05:29:37.049584 32403 solver.cpp:406] Test net output #112: loss3/accuracy14 = 0.958
I0607 05:29:37.049604 32403 solver.cpp:406] Test net output #113: loss3/accuracy15 = 0.964
I0607 05:29:37.049631 32403 solver.cpp:406] Test net output #114: loss3/accuracy16 = 0.97
I0607 05:29:37.049670 32403 solver.cpp:406] Test net output #115: loss3/accuracy17 = 0.988
I0607 05:29:37.049695 32403 solver.cpp:406] Test net output #116: loss3/accuracy18 = 0.989
I0607 05:29:37.049716 32403 solver.cpp:406] Test net output #117: loss3/accuracy19 = 0.991
I0607 05:29:37.049736 32403 solver.cpp:406] Test net output #118: loss3/accuracy20 = 0.995
I0607 05:29:37.049757 32403 solver.cpp:406] Test net output #119: loss3/accuracy21 = 0.998
I0607 05:29:37.049777 32403 solver.cpp:406] Test net output #120: loss3/accuracy22 = 1
I0607 05:29:37.049796 32403 solver.cpp:406] Test net output #121: loss3/accuracy_incl_empty = 0.9195
I0607 05:29:37.049816 32403 solver.cpp:406] Test net output #122: loss3/accuracy_top3 = 0.90775
I0607 05:29:37.049840 32403 solver.cpp:406] Test net output #123: loss3/cross_entropy_loss = 0.834818 (* 1 = 0.834818 loss)
I0607 05:29:37.049872 32403 solver.cpp:406] Test net output #124: loss3/cross_entropy_loss_incl_empty = 0.335915 (* 1 = 0.335915 loss)
I0607 05:29:37.049897 32403 solver.cpp:406] Test net output #125: loss3/loss01 = 0.602414 (* 0.0909091 = 0.0547649 loss)
I0607 05:29:37.049923 32403 solver.cpp:406] Test net output #126: loss3/loss02 = 0.770355 (* 0.0909091 = 0.0700322 loss)
I0607 05:29:37.049948 32403 solver.cpp:406] Test net output #127: loss3/loss03 = 0.734562 (* 0.0909091 = 0.0667784 loss)
I0607 05:29:37.049973 32403 solver.cpp:406] Test net output #128: loss3/loss04 = 0.763181 (* 0.0909091 = 0.0693801 loss)
I0607 05:29:37.050006 32403 solver.cpp:406] Test net output #129: loss3/loss05 = 0.778224 (* 0.0909091 = 0.0707476 loss)
I0607 05:29:37.050030 32403 solver.cpp:406] Test net output #130: loss3/loss06 = 0.687261 (* 0.0909091 = 0.0624782 loss)
I0607 05:29:37.050055 32403 solver.cpp:406] Test net output #131: loss3/loss07 = 0.543564 (* 0.0909091 = 0.0494149 loss)
I0607 05:29:37.050077 32403 solver.cpp:406] Test net output #132: loss3/loss08 = 0.489798 (* 0.0909091 = 0.0445271 loss)
I0607 05:29:37.050101 32403 solver.cpp:406] Test net output #133: loss3/loss09 = 0.404629 (* 0.0909091 = 0.0367845 loss)
I0607 05:29:37.050125 32403 solver.cpp:406] Test net output #134: loss3/loss10 = 0.376624 (* 0.0909091 = 0.0342386 loss)
I0607 05:29:37.050151 32403 solver.cpp:406] Test net output #135: loss3/loss11 = 0.293002 (* 0.0909091 = 0.0266365 loss)
I0607 05:29:37.050176 32403 solver.cpp:406] Test net output #136: loss3/loss12 = 0.268941 (* 0.0909091 = 0.0244492 loss)
I0607 05:29:37.050196 32403 solver.cpp:406] Test net output #137: loss3/loss13 = 0.190657 (* 0.0909091 = 0.0173325 loss)
I0607 05:29:37.050217 32403 solver.cpp:406] Test net output #138: loss3/loss14 = 0.144657 (* 0.0909091 = 0.0131506 loss)
I0607 05:29:37.050241 32403 solver.cpp:406] Test net output #139: loss3/loss15 = 0.125714 (* 0.0909091 = 0.0114285 loss)
I0607 05:29:37.050266 32403 solver.cpp:406] Test net output #140: loss3/loss16 = 0.107699 (* 0.0909091 = 0.0097908 loss)
I0607 05:29:37.050292 32403 solver.cpp:406] Test net output #141: loss3/loss17 = 0.0572213 (* 0.0909091 = 0.00520194 loss)
I0607 05:29:37.050315 32403 solver.cpp:406] Test net output #142: loss3/loss18 = 0.041725 (* 0.0909091 = 0.00379319 loss)
I0607 05:29:37.050338 32403 solver.cpp:406] Test net output #143: loss3/loss19 = 0.0360564 (* 0.0909091 = 0.00327785 loss)
I0607 05:29:37.050362 32403 solver.cpp:406] Test net output #144: loss3/loss20 = 0.0192199 (* 0.0909091 = 0.00174727 loss)
I0607 05:29:37.050387 32403 solver.cpp:406] Test net output #145: loss3/loss21 = 0.00931644 (* 0.0909091 = 0.000846949 loss)
I0607 05:29:37.050412 32403 solver.cpp:406] Test net output #146: loss3/loss22 = 0.00109285 (* 0.0909091 = 9.93501e-05 loss)
I0607 05:29:37.050432 32403 solver.cpp:406] Test net output #147: total_accuracy = 0.505
I0607 05:29:37.050452 32403 solver.cpp:406] Test net output #148: total_accuracy_not_rec = 0.496
I0607 05:29:37.050472 32403 solver.cpp:406] Test net output #149: total_confidence = 0.424579
I0607 05:29:37.050508 32403 solver.cpp:406] Test net output #150: total_confidence_not_rec = 0.393119
I0607 05:29:37.407459 32403 solver.cpp:229] Iteration 20000, loss = 3.93866
I0607 05:29:37.407524 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.517241
I0607 05:29:37.407553 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 1
I0607 05:29:37.407577 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.875
I0607 05:29:37.407599 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.625
I0607 05:29:37.407621 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.625
I0607 05:29:37.407644 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.625
I0607 05:29:37.407666 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.625
I0607 05:29:37.407686 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.875
I0607 05:29:37.407711 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.875
I0607 05:29:37.407732 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.875
I0607 05:29:37.407754 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.75
I0607 05:29:37.407775 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.75
I0607 05:29:37.407796 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.875
I0607 05:29:37.407819 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.875
I0607 05:29:37.407841 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 1
I0607 05:29:37.407862 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 05:29:37.407882 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 05:29:37.407904 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 05:29:37.407924 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 05:29:37.407944 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 05:29:37.407963 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 05:29:37.407984 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 05:29:37.408004 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 05:29:37.408025 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.835227
I0607 05:29:37.408046 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.775862
I0607 05:29:37.408072 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.54087 (* 0.3 = 0.46226 loss)
I0607 05:29:37.408098 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.543077 (* 0.3 = 0.162923 loss)
I0607 05:29:37.408124 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.364084 (* 0.0272727 = 0.00992956 loss)
I0607 05:29:37.408151 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 0.826406 (* 0.0272727 = 0.0225383 loss)
I0607 05:29:37.408179 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.26708 (* 0.0272727 = 0.0345568 loss)
I0607 05:29:37.408205 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.12692 (* 0.0272727 = 0.0307343 loss)
I0607 05:29:37.408234 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.14347 (* 0.0272727 = 0.0311855 loss)
I0607 05:29:37.408260 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.22936 (* 0.0272727 = 0.033528 loss)
I0607 05:29:37.408284 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 0.440805 (* 0.0272727 = 0.012022 loss)
I0607 05:29:37.408309 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.971624 (* 0.0272727 = 0.0264989 loss)
I0607 05:29:37.408334 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.635428 (* 0.0272727 = 0.0173298 loss)
I0607 05:29:37.408359 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.645049 (* 0.0272727 = 0.0175923 loss)
I0607 05:29:37.408382 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 1.27578 (* 0.0272727 = 0.034794 loss)
I0607 05:29:37.408445 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.389044 (* 0.0272727 = 0.0106103 loss)
I0607 05:29:37.408473 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.668383 (* 0.0272727 = 0.0182286 loss)
I0607 05:29:37.408499 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.00272182 (* 0.0272727 = 7.42314e-05 loss)
I0607 05:29:37.408524 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.000514013 (* 0.0272727 = 1.40185e-05 loss)
I0607 05:29:37.408550 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.000160325 (* 0.0272727 = 4.37251e-06 loss)
I0607 05:29:37.408576 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 4.48487e-05 (* 0.0272727 = 1.22315e-06 loss)
I0607 05:29:37.408601 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 4.32249e-05 (* 0.0272727 = 1.17886e-06 loss)
I0607 05:29:37.408627 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 2.41046e-05 (* 0.0272727 = 6.57399e-07 loss)
I0607 05:29:37.408653 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 5.31976e-06 (* 0.0272727 = 1.45084e-07 loss)
I0607 05:29:37.408677 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 3.03986e-06 (* 0.0272727 = 8.29054e-08 loss)
I0607 05:29:37.408701 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 4.75355e-06 (* 0.0272727 = 1.29642e-07 loss)
I0607 05:29:37.408721 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.62069
I0607 05:29:37.408745 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.875
I0607 05:29:37.408771 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.875
I0607 05:29:37.408793 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.875
I0607 05:29:37.408814 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.75
I0607 05:29:37.408835 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.625
I0607 05:29:37.408855 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.75
I0607 05:29:37.408876 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.875
I0607 05:29:37.408896 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.75
I0607 05:29:37.408917 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.75
I0607 05:29:37.408938 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 1
I0607 05:29:37.408958 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.75
I0607 05:29:37.408979 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.875
I0607 05:29:37.408999 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.875
I0607 05:29:37.409020 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 1
I0607 05:29:37.409040 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 05:29:37.409061 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 05:29:37.409081 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 05:29:37.409102 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 05:29:37.409142 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 05:29:37.409168 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 05:29:37.409189 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 05:29:37.409210 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 05:29:37.409231 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.857955
I0607 05:29:37.409252 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.862069
I0607 05:29:37.409283 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.08684 (* 0.3 = 0.326051 loss)
I0607 05:29:37.409309 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.411325 (* 0.3 = 0.123397 loss)
I0607 05:29:37.409354 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.32423 (* 0.0272727 = 0.00884264 loss)
I0607 05:29:37.409381 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.234821 (* 0.0272727 = 0.0064042 loss)
I0607 05:29:37.409405 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.710908 (* 0.0272727 = 0.0193884 loss)
I0607 05:29:37.409431 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 0.64427 (* 0.0272727 = 0.017571 loss)
I0607 05:29:37.409456 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 0.723873 (* 0.0272727 = 0.019742 loss)
I0607 05:29:37.409481 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.47361 (* 0.0272727 = 0.0401893 loss)
I0607 05:29:37.409505 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 0.471118 (* 0.0272727 = 0.0128487 loss)
I0607 05:29:37.409531 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.835071 (* 0.0272727 = 0.0227747 loss)
I0607 05:29:37.409556 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.623008 (* 0.0272727 = 0.0169911 loss)
I0607 05:29:37.409581 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.352506 (* 0.0272727 = 0.00961379 loss)
I0607 05:29:37.409607 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.740392 (* 0.0272727 = 0.0201925 loss)
I0607 05:29:37.409636 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 0.507847 (* 0.0272727 = 0.0138504 loss)
I0607 05:29:37.409662 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.50945 (* 0.0272727 = 0.0138941 loss)
I0607 05:29:37.409688 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 0.0391827 (* 0.0272727 = 0.00106862 loss)
I0607 05:29:37.409713 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.00939452 (* 0.0272727 = 0.000256214 loss)
I0607 05:29:37.409739 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.00272634 (* 0.0272727 = 7.43547e-05 loss)
I0607 05:29:37.409765 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.000605298 (* 0.0272727 = 1.65081e-05 loss)
I0607 05:29:37.409792 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.000162656 (* 0.0272727 = 4.43607e-06 loss)
I0607 05:29:37.409822 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.000157598 (* 0.0272727 = 4.29812e-06 loss)
I0607 05:29:37.409849 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 0.000241807 (* 0.0272727 = 6.59472e-06 loss)
I0607 05:29:37.409876 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 0.000105423 (* 0.0272727 = 2.87518e-06 loss)
I0607 05:29:37.409903 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 8.54674e-05 (* 0.0272727 = 2.33093e-06 loss)
I0607 05:29:37.409924 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.948276
I0607 05:29:37.409946 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 05:29:37.409968 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 1
I0607 05:29:37.409991 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 1
I0607 05:29:37.410012 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 1
I0607 05:29:37.410032 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 1
I0607 05:29:37.410054 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.875
I0607 05:29:37.410076 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 1
I0607 05:29:37.410097 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.875
I0607 05:29:37.410120 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 1
I0607 05:29:37.410141 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.875
I0607 05:29:37.410162 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.75
I0607 05:29:37.410183 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.875
I0607 05:29:37.410204 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 1
I0607 05:29:37.410240 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 1
I0607 05:29:37.410264 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 05:29:37.410284 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 05:29:37.410306 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 05:29:37.410332 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 05:29:37.410353 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 05:29:37.410373 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 05:29:37.410393 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 05:29:37.410415 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 05:29:37.410435 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.977273
I0607 05:29:37.410456 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.982759
I0607 05:29:37.410481 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.249084 (* 1 = 0.249084 loss)
I0607 05:29:37.410506 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.0989358 (* 1 = 0.0989358 loss)
I0607 05:29:37.410531 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.0228098 (* 0.0909091 = 0.00207362 loss)
I0607 05:29:37.410557 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.0494525 (* 0.0909091 = 0.00449568 loss)
I0607 05:29:37.410583 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.151127 (* 0.0909091 = 0.0137388 loss)
I0607 05:29:37.410606 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.106628 (* 0.0909091 = 0.00969347 loss)
I0607 05:29:37.410632 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.202199 (* 0.0909091 = 0.0183817 loss)
I0607 05:29:37.410658 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.510536 (* 0.0909091 = 0.0464124 loss)
I0607 05:29:37.410684 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.136522 (* 0.0909091 = 0.0124111 loss)
I0607 05:29:37.410709 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.428487 (* 0.0909091 = 0.0389534 loss)
I0607 05:29:37.410734 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.167681 (* 0.0909091 = 0.0152437 loss)
I0607 05:29:37.410759 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 0.296187 (* 0.0909091 = 0.0269261 loss)
I0607 05:29:37.410784 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 0.6001 (* 0.0909091 = 0.0545545 loss)
I0607 05:29:37.410809 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 0.218315 (* 0.0909091 = 0.0198468 loss)
I0607 05:29:37.410830 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.0989698 (* 0.0909091 = 0.00899725 loss)
I0607 05:29:37.410862 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 0.00990293 (* 0.0909091 = 0.000900266 loss)
I0607 05:29:37.410889 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.00407729 (* 0.0909091 = 0.000370663 loss)
I0607 05:29:37.410914 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.000968307 (* 0.0909091 = 8.80279e-05 loss)
I0607 05:29:37.410940 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.000322325 (* 0.0909091 = 2.93023e-05 loss)
I0607 05:29:37.410966 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.000203624 (* 0.0909091 = 1.85112e-05 loss)
I0607 05:29:37.410991 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.000193082 (* 0.0909091 = 1.75529e-05 loss)
I0607 05:29:37.411016 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.000159028 (* 0.0909091 = 1.44571e-05 loss)
I0607 05:29:37.411042 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.000138368 (* 0.0909091 = 1.25789e-05 loss)
I0607 05:29:37.411068 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 0.000109898 (* 0.0909091 = 9.99075e-06 loss)
I0607 05:29:37.411105 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.5
I0607 05:29:37.411129 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.625
I0607 05:29:37.411150 32403 solver.cpp:245] Train net output #149: total_confidence = 0.400926
I0607 05:29:37.411171 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.374267
I0607 05:29:37.411192 32403 sgd_solver.cpp:106] Iteration 20000, lr = 0.001
I0607 05:29:44.713094 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 38.128 > 30) by scale factor 0.786824
I0607 05:30:30.298918 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 45.5589 > 30) by scale factor 0.658488
I0607 05:31:23.591320 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 51.1623 > 30) by scale factor 0.586369
I0607 05:31:55.294358 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 34.577 > 30) by scale factor 0.86763
I0607 05:32:43.919740 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.1012 > 30) by scale factor 0.996639
I0607 05:33:49.576864 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 33.6259 > 30) by scale factor 0.892168
I0607 05:34:07.312538 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 38.9804 > 30) by scale factor 0.769617
I0607 05:34:18.907696 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 35.1057 > 30) by scale factor 0.854563
I0607 05:35:14.515059 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.8965 > 30) by scale factor 0.940543
I0607 05:35:33.029865 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 37.0449 > 30) by scale factor 0.809829
I0607 05:36:03.542973 32403 solver.cpp:229] Iteration 20500, loss = 3.89044
I0607 05:36:03.543121 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.477612
I0607 05:36:03.543143 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.875
I0607 05:36:03.543157 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.5
I0607 05:36:03.543170 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.625
I0607 05:36:03.543184 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.25
I0607 05:36:03.543196 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.25
I0607 05:36:03.543210 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.625
I0607 05:36:03.543222 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.375
I0607 05:36:03.543236 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.875
I0607 05:36:03.543248 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 0.75
I0607 05:36:03.543262 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 0.625
I0607 05:36:03.543274 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 0.625
I0607 05:36:03.543287 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 0.75
I0607 05:36:03.543299 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 0.75
I0607 05:36:03.543311 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 0.75
I0607 05:36:03.543324 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 05:36:03.543336 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 05:36:03.543349 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 05:36:03.543360 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 05:36:03.543372 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 05:36:03.543385 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 05:36:03.543397 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 05:36:03.543409 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 05:36:03.543421 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.784091
I0607 05:36:03.543434 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.716418
I0607 05:36:03.543452 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 1.9034 (* 0.3 = 0.571021 loss)
I0607 05:36:03.543467 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.788441 (* 0.3 = 0.236532 loss)
I0607 05:36:03.543481 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 0.251227 (* 0.0272727 = 0.00685164 loss)
I0607 05:36:03.543496 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.50823 (* 0.0272727 = 0.0411335 loss)
I0607 05:36:03.543511 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 1.20202 (* 0.0272727 = 0.0327824 loss)
I0607 05:36:03.543525 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 1.94135 (* 0.0272727 = 0.0529459 loss)
I0607 05:36:03.543540 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 1.79397 (* 0.0272727 = 0.0489264 loss)
I0607 05:36:03.543555 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.6895 (* 0.0272727 = 0.0460772 loss)
I0607 05:36:03.543568 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 1.21555 (* 0.0272727 = 0.0331513 loss)
I0607 05:36:03.543582 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 1.16039 (* 0.0272727 = 0.0316469 loss)
I0607 05:36:03.543597 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 1.07577 (* 0.0272727 = 0.0293392 loss)
I0607 05:36:03.543612 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 1.54048 (* 0.0272727 = 0.0420131 loss)
I0607 05:36:03.543625 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 1.44224 (* 0.0272727 = 0.0393339 loss)
I0607 05:36:03.543639 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 1.85947 (* 0.0272727 = 0.0507127 loss)
I0607 05:36:03.543675 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.911578 (* 0.0272727 = 0.0248612 loss)
I0607 05:36:03.543691 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 1.72013 (* 0.0272727 = 0.0469127 loss)
I0607 05:36:03.543706 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.0213779 (* 0.0272727 = 0.000583032 loss)
I0607 05:36:03.543720 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.00420301 (* 0.0272727 = 0.000114628 loss)
I0607 05:36:03.543735 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.00188696 (* 0.0272727 = 5.14624e-05 loss)
I0607 05:36:03.543750 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.00029443 (* 0.0272727 = 8.0299e-06 loss)
I0607 05:36:03.543764 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 5.93932e-05 (* 0.0272727 = 1.61981e-06 loss)
I0607 05:36:03.543778 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 5.35213e-05 (* 0.0272727 = 1.45967e-06 loss)
I0607 05:36:03.543793 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 1.71964e-05 (* 0.0272727 = 4.68993e-07 loss)
I0607 05:36:03.543808 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 1.13995e-05 (* 0.0272727 = 3.10896e-07 loss)
I0607 05:36:03.543820 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.567164
I0607 05:36:03.543833 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 1
I0607 05:36:03.543845 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.875
I0607 05:36:03.543858 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 1
I0607 05:36:03.543869 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.875
I0607 05:36:03.543884 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.375
I0607 05:36:03.543897 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.375
I0607 05:36:03.543910 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.625
I0607 05:36:03.543922 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.625
I0607 05:36:03.543934 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.75
I0607 05:36:03.543946 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 0.625
I0607 05:36:03.543958 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 0.625
I0607 05:36:03.543970 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 0.75
I0607 05:36:03.543982 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 0.75
I0607 05:36:03.543994 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 0.75
I0607 05:36:03.544008 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 05:36:03.544019 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 05:36:03.544031 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 05:36:03.544044 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 05:36:03.544055 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 05:36:03.544067 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 05:36:03.544080 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 05:36:03.544091 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 05:36:03.544103 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.829545
I0607 05:36:03.544116 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.776119
I0607 05:36:03.544129 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.70373 (* 0.3 = 0.511119 loss)
I0607 05:36:03.544148 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.683406 (* 0.3 = 0.205022 loss)
I0607 05:36:03.544162 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 0.100086 (* 0.0272727 = 0.00272962 loss)
I0607 05:36:03.544178 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 0.39907 (* 0.0272727 = 0.0108837 loss)
I0607 05:36:03.544203 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 0.638847 (* 0.0272727 = 0.0174231 loss)
I0607 05:36:03.544219 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 0.709006 (* 0.0272727 = 0.0193365 loss)
I0607 05:36:03.544234 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 1.47738 (* 0.0272727 = 0.0402923 loss)
I0607 05:36:03.544247 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.52846 (* 0.0272727 = 0.0416853 loss)
I0607 05:36:03.544261 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 1.23909 (* 0.0272727 = 0.0337934 loss)
I0607 05:36:03.544276 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 1.15897 (* 0.0272727 = 0.0316082 loss)
I0607 05:36:03.544291 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.830611 (* 0.0272727 = 0.022653 loss)
I0607 05:36:03.544304 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 1.80281 (* 0.0272727 = 0.0491676 loss)
I0607 05:36:03.544318 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 1.55295 (* 0.0272727 = 0.0423533 loss)
I0607 05:36:03.544332 32403 solver.cpp:245] Train net output #87: loss2/loss12 = 2.03547 (* 0.0272727 = 0.0555128 loss)
I0607 05:36:03.544347 32403 solver.cpp:245] Train net output #88: loss2/loss13 = 0.605647 (* 0.0272727 = 0.0165177 loss)
I0607 05:36:03.544363 32403 solver.cpp:245] Train net output #89: loss2/loss14 = 1.62198 (* 0.0272727 = 0.0442358 loss)
I0607 05:36:03.544376 32403 solver.cpp:245] Train net output #90: loss2/loss15 = 0.0229396 (* 0.0272727 = 0.000625626 loss)
I0607 05:36:03.544390 32403 solver.cpp:245] Train net output #91: loss2/loss16 = 0.00812428 (* 0.0272727 = 0.000221571 loss)
I0607 05:36:03.544405 32403 solver.cpp:245] Train net output #92: loss2/loss17 = 0.00453166 (* 0.0272727 = 0.000123591 loss)
I0607 05:36:03.544420 32403 solver.cpp:245] Train net output #93: loss2/loss18 = 0.000764679 (* 0.0272727 = 2.08549e-05 loss)
I0607 05:36:03.544435 32403 solver.cpp:245] Train net output #94: loss2/loss19 = 0.00019545 (* 0.0272727 = 5.33047e-06 loss)
I0607 05:36:03.544448 32403 solver.cpp:245] Train net output #95: loss2/loss20 = 5.64882e-05 (* 0.0272727 = 1.54059e-06 loss)
I0607 05:36:03.544463 32403 solver.cpp:245] Train net output #96: loss2/loss21 = 6.38614e-05 (* 0.0272727 = 1.74168e-06 loss)
I0607 05:36:03.544477 32403 solver.cpp:245] Train net output #97: loss2/loss22 = 1.70473e-05 (* 0.0272727 = 4.64927e-07 loss)
I0607 05:36:03.544489 32403 solver.cpp:245] Train net output #98: loss3/accuracy = 0.791045
I0607 05:36:03.544502 32403 solver.cpp:245] Train net output #99: loss3/accuracy01 = 1
I0607 05:36:03.544514 32403 solver.cpp:245] Train net output #100: loss3/accuracy02 = 1
I0607 05:36:03.544526 32403 solver.cpp:245] Train net output #101: loss3/accuracy03 = 1
I0607 05:36:03.544538 32403 solver.cpp:245] Train net output #102: loss3/accuracy04 = 1
I0607 05:36:03.544550 32403 solver.cpp:245] Train net output #103: loss3/accuracy05 = 1
I0607 05:36:03.544562 32403 solver.cpp:245] Train net output #104: loss3/accuracy06 = 0.875
I0607 05:36:03.544574 32403 solver.cpp:245] Train net output #105: loss3/accuracy07 = 0.875
I0607 05:36:03.544586 32403 solver.cpp:245] Train net output #106: loss3/accuracy08 = 0.875
I0607 05:36:03.544598 32403 solver.cpp:245] Train net output #107: loss3/accuracy09 = 0.75
I0607 05:36:03.544610 32403 solver.cpp:245] Train net output #108: loss3/accuracy10 = 0.75
I0607 05:36:03.544623 32403 solver.cpp:245] Train net output #109: loss3/accuracy11 = 0.75
I0607 05:36:03.544636 32403 solver.cpp:245] Train net output #110: loss3/accuracy12 = 0.875
I0607 05:36:03.544647 32403 solver.cpp:245] Train net output #111: loss3/accuracy13 = 0.875
I0607 05:36:03.544659 32403 solver.cpp:245] Train net output #112: loss3/accuracy14 = 0.75
I0607 05:36:03.544672 32403 solver.cpp:245] Train net output #113: loss3/accuracy15 = 1
I0607 05:36:03.544683 32403 solver.cpp:245] Train net output #114: loss3/accuracy16 = 1
I0607 05:36:03.544705 32403 solver.cpp:245] Train net output #115: loss3/accuracy17 = 1
I0607 05:36:03.544719 32403 solver.cpp:245] Train net output #116: loss3/accuracy18 = 1
I0607 05:36:03.544731 32403 solver.cpp:245] Train net output #117: loss3/accuracy19 = 1
I0607 05:36:03.544744 32403 solver.cpp:245] Train net output #118: loss3/accuracy20 = 1
I0607 05:36:03.544755 32403 solver.cpp:245] Train net output #119: loss3/accuracy21 = 1
I0607 05:36:03.544767 32403 solver.cpp:245] Train net output #120: loss3/accuracy22 = 1
I0607 05:36:03.544778 32403 solver.cpp:245] Train net output #121: loss3/accuracy_incl_empty = 0.903409
I0607 05:36:03.544791 32403 solver.cpp:245] Train net output #122: loss3/accuracy_top3 = 0.850746
I0607 05:36:03.544806 32403 solver.cpp:245] Train net output #123: loss3/cross_entropy_loss = 0.997346 (* 1 = 0.997346 loss)
I0607 05:36:03.544821 32403 solver.cpp:245] Train net output #124: loss3/cross_entropy_loss_incl_empty = 0.422044 (* 1 = 0.422044 loss)
I0607 05:36:03.544834 32403 solver.cpp:245] Train net output #125: loss3/loss01 = 0.0196106 (* 0.0909091 = 0.00178278 loss)
I0607 05:36:03.544849 32403 solver.cpp:245] Train net output #126: loss3/loss02 = 0.03697 (* 0.0909091 = 0.00336091 loss)
I0607 05:36:03.544864 32403 solver.cpp:245] Train net output #127: loss3/loss03 = 0.0429259 (* 0.0909091 = 0.00390236 loss)
I0607 05:36:03.544878 32403 solver.cpp:245] Train net output #128: loss3/loss04 = 0.105496 (* 0.0909091 = 0.00959052 loss)
I0607 05:36:03.544893 32403 solver.cpp:245] Train net output #129: loss3/loss05 = 0.0834776 (* 0.0909091 = 0.00758888 loss)
I0607 05:36:03.544908 32403 solver.cpp:245] Train net output #130: loss3/loss06 = 0.702985 (* 0.0909091 = 0.0639077 loss)
I0607 05:36:03.544921 32403 solver.cpp:245] Train net output #131: loss3/loss07 = 0.980056 (* 0.0909091 = 0.089096 loss)
I0607 05:36:03.544939 32403 solver.cpp:245] Train net output #132: loss3/loss08 = 0.971345 (* 0.0909091 = 0.0883041 loss)
I0607 05:36:03.544953 32403 solver.cpp:245] Train net output #133: loss3/loss09 = 0.946334 (* 0.0909091 = 0.0860304 loss)
I0607 05:36:03.544967 32403 solver.cpp:245] Train net output #134: loss3/loss10 = 1.1664 (* 0.0909091 = 0.106037 loss)
I0607 05:36:03.544981 32403 solver.cpp:245] Train net output #135: loss3/loss11 = 1.01302 (* 0.0909091 = 0.0920923 loss)
I0607 05:36:03.544996 32403 solver.cpp:245] Train net output #136: loss3/loss12 = 1.1031 (* 0.0909091 = 0.100282 loss)
I0607 05:36:03.545009 32403 solver.cpp:245] Train net output #137: loss3/loss13 = 0.302693 (* 0.0909091 = 0.0275176 loss)
I0607 05:36:03.545024 32403 solver.cpp:245] Train net output #138: loss3/loss14 = 1.24073 (* 0.0909091 = 0.112793 loss)
I0607 05:36:03.545038 32403 solver.cpp:245] Train net output #139: loss3/loss15 = 0.027222 (* 0.0909091 = 0.00247473 loss)
I0607 05:36:03.545053 32403 solver.cpp:245] Train net output #140: loss3/loss16 = 0.0108612 (* 0.0909091 = 0.000987382 loss)
I0607 05:36:03.545068 32403 solver.cpp:245] Train net output #141: loss3/loss17 = 0.00294855 (* 0.0909091 = 0.00026805 loss)
I0607 05:36:03.545078 32403 solver.cpp:245] Train net output #142: loss3/loss18 = 0.000645825 (* 0.0909091 = 5.87114e-05 loss)
I0607 05:36:03.545089 32403 solver.cpp:245] Train net output #143: loss3/loss19 = 0.000476172 (* 0.0909091 = 4.32883e-05 loss)
I0607 05:36:03.545104 32403 solver.cpp:245] Train net output #144: loss3/loss20 = 0.000230608 (* 0.0909091 = 2.09643e-05 loss)
I0607 05:36:03.545130 32403 solver.cpp:245] Train net output #145: loss3/loss21 = 0.000201323 (* 0.0909091 = 1.83021e-05 loss)
I0607 05:36:03.545147 32403 solver.cpp:245] Train net output #146: loss3/loss22 = 8.38002e-05 (* 0.0909091 = 7.6182e-06 loss)
I0607 05:36:03.545161 32403 solver.cpp:245] Train net output #147: total_accuracy = 0.5
I0607 05:36:03.545172 32403 solver.cpp:245] Train net output #148: total_accuracy_not_rec = 0.625
I0607 05:36:03.545200 32403 solver.cpp:245] Train net output #149: total_confidence = 0.419882
I0607 05:36:03.545214 32403 solver.cpp:245] Train net output #150: total_confidence_not_rec = 0.417604
I0607 05:36:03.545228 32403 sgd_solver.cpp:106] Iteration 20500, lr = 0.001
I0607 05:36:12.383311 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 32.7704 > 30) by scale factor 0.91546
I0607 05:36:24.757169 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 31.8708 > 30) by scale factor 0.941299
I0607 05:37:35.057796 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.7438 > 30) by scale factor 0.975807
I0607 05:39:23.158931 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 43.7057 > 30) by scale factor 0.686409
I0607 05:40:01.736199 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 42.8356 > 30) by scale factor 0.700351
I0607 05:41:36.683344 32403 sgd_solver.cpp:92] Gradient clipping: scaling down gradients (L2 norm 30.5342 > 30) by scale factor 0.982504
I0607 05:42:29.586789 32403 solver.cpp:229] Iteration 21000, loss = 3.93589
I0607 05:42:29.586915 32403 solver.cpp:245] Train net output #0: loss1/accuracy = 0.466667
I0607 05:42:29.586936 32403 solver.cpp:245] Train net output #1: loss1/accuracy01 = 0.5
I0607 05:42:29.586951 32403 solver.cpp:245] Train net output #2: loss1/accuracy02 = 0.5
I0607 05:42:29.586966 32403 solver.cpp:245] Train net output #3: loss1/accuracy03 = 0.5
I0607 05:42:29.586977 32403 solver.cpp:245] Train net output #4: loss1/accuracy04 = 0.25
I0607 05:42:29.586990 32403 solver.cpp:245] Train net output #5: loss1/accuracy05 = 0.375
I0607 05:42:29.587003 32403 solver.cpp:245] Train net output #6: loss1/accuracy06 = 0.625
I0607 05:42:29.587016 32403 solver.cpp:245] Train net output #7: loss1/accuracy07 = 0.75
I0607 05:42:29.587031 32403 solver.cpp:245] Train net output #8: loss1/accuracy08 = 0.75
I0607 05:42:29.587044 32403 solver.cpp:245] Train net output #9: loss1/accuracy09 = 1
I0607 05:42:29.587057 32403 solver.cpp:245] Train net output #10: loss1/accuracy10 = 1
I0607 05:42:29.587069 32403 solver.cpp:245] Train net output #11: loss1/accuracy11 = 1
I0607 05:42:29.587081 32403 solver.cpp:245] Train net output #12: loss1/accuracy12 = 1
I0607 05:42:29.587095 32403 solver.cpp:245] Train net output #13: loss1/accuracy13 = 1
I0607 05:42:29.587106 32403 solver.cpp:245] Train net output #14: loss1/accuracy14 = 1
I0607 05:42:29.587118 32403 solver.cpp:245] Train net output #15: loss1/accuracy15 = 1
I0607 05:42:29.587131 32403 solver.cpp:245] Train net output #16: loss1/accuracy16 = 1
I0607 05:42:29.587142 32403 solver.cpp:245] Train net output #17: loss1/accuracy17 = 1
I0607 05:42:29.587154 32403 solver.cpp:245] Train net output #18: loss1/accuracy18 = 1
I0607 05:42:29.587167 32403 solver.cpp:245] Train net output #19: loss1/accuracy19 = 1
I0607 05:42:29.587178 32403 solver.cpp:245] Train net output #20: loss1/accuracy20 = 1
I0607 05:42:29.587190 32403 solver.cpp:245] Train net output #21: loss1/accuracy21 = 1
I0607 05:42:29.587203 32403 solver.cpp:245] Train net output #22: loss1/accuracy22 = 1
I0607 05:42:29.587214 32403 solver.cpp:245] Train net output #23: loss1/accuracy_incl_empty = 0.846591
I0607 05:42:29.587227 32403 solver.cpp:245] Train net output #24: loss1/accuracy_top3 = 0.711111
I0607 05:42:29.587244 32403 solver.cpp:245] Train net output #25: loss1/cross_entropy_loss = 2.14626 (* 0.3 = 0.643878 loss)
I0607 05:42:29.587258 32403 solver.cpp:245] Train net output #26: loss1/cross_entropy_loss_incl_empty = 0.625239 (* 0.3 = 0.187572 loss)
I0607 05:42:29.587273 32403 solver.cpp:245] Train net output #27: loss1/loss01 = 1.3696 (* 0.0272727 = 0.0373526 loss)
I0607 05:42:29.587287 32403 solver.cpp:245] Train net output #28: loss1/loss02 = 1.85679 (* 0.0272727 = 0.0506397 loss)
I0607 05:42:29.587301 32403 solver.cpp:245] Train net output #29: loss1/loss03 = 3.10825 (* 0.0272727 = 0.0847705 loss)
I0607 05:42:29.587316 32403 solver.cpp:245] Train net output #30: loss1/loss04 = 2.21248 (* 0.0272727 = 0.0603403 loss)
I0607 05:42:29.587330 32403 solver.cpp:245] Train net output #31: loss1/loss05 = 2.21377 (* 0.0272727 = 0.0603756 loss)
I0607 05:42:29.587344 32403 solver.cpp:245] Train net output #32: loss1/loss06 = 1.14224 (* 0.0272727 = 0.0311519 loss)
I0607 05:42:29.587358 32403 solver.cpp:245] Train net output #33: loss1/loss07 = 1.52897 (* 0.0272727 = 0.0416992 loss)
I0607 05:42:29.587373 32403 solver.cpp:245] Train net output #34: loss1/loss08 = 0.816677 (* 0.0272727 = 0.022273 loss)
I0607 05:42:29.587388 32403 solver.cpp:245] Train net output #35: loss1/loss09 = 0.161892 (* 0.0272727 = 0.00441524 loss)
I0607 05:42:29.587402 32403 solver.cpp:245] Train net output #36: loss1/loss10 = 0.0673542 (* 0.0272727 = 0.00183693 loss)
I0607 05:42:29.587416 32403 solver.cpp:245] Train net output #37: loss1/loss11 = 0.0269861 (* 0.0272727 = 0.000735985 loss)
I0607 05:42:29.587430 32403 solver.cpp:245] Train net output #38: loss1/loss12 = 0.00773529 (* 0.0272727 = 0.000210963 loss)
I0607 05:42:29.587445 32403 solver.cpp:245] Train net output #39: loss1/loss13 = 0.00474189 (* 0.0272727 = 0.000129324 loss)
I0607 05:42:29.587479 32403 solver.cpp:245] Train net output #40: loss1/loss14 = 0.00346188 (* 0.0272727 = 9.44148e-05 loss)
I0607 05:42:29.587496 32403 solver.cpp:245] Train net output #41: loss1/loss15 = 0.00167782 (* 0.0272727 = 4.57587e-05 loss)
I0607 05:42:29.587510 32403 solver.cpp:245] Train net output #42: loss1/loss16 = 0.0017352 (* 0.0272727 = 4.73237e-05 loss)
I0607 05:42:29.587524 32403 solver.cpp:245] Train net output #43: loss1/loss17 = 0.00189508 (* 0.0272727 = 5.1684e-05 loss)
I0607 05:42:29.587540 32403 solver.cpp:245] Train net output #44: loss1/loss18 = 0.00227047 (* 0.0272727 = 6.19218e-05 loss)
I0607 05:42:29.587554 32403 solver.cpp:245] Train net output #45: loss1/loss19 = 0.00167998 (* 0.0272727 = 4.58177e-05 loss)
I0607 05:42:29.587569 32403 solver.cpp:245] Train net output #46: loss1/loss20 = 0.00304779 (* 0.0272727 = 8.31214e-05 loss)
I0607 05:42:29.587584 32403 solver.cpp:245] Train net output #47: loss1/loss21 = 0.00182458 (* 0.0272727 = 4.97612e-05 loss)
I0607 05:42:29.587597 32403 solver.cpp:245] Train net output #48: loss1/loss22 = 0.00351027 (* 0.0272727 = 9.57345e-05 loss)
I0607 05:42:29.587610 32403 solver.cpp:245] Train net output #49: loss2/accuracy = 0.622222
I0607 05:42:29.587623 32403 solver.cpp:245] Train net output #50: loss2/accuracy01 = 0.75
I0607 05:42:29.587635 32403 solver.cpp:245] Train net output #51: loss2/accuracy02 = 0.625
I0607 05:42:29.587647 32403 solver.cpp:245] Train net output #52: loss2/accuracy03 = 0.625
I0607 05:42:29.587659 32403 solver.cpp:245] Train net output #53: loss2/accuracy04 = 0.625
I0607 05:42:29.587671 32403 solver.cpp:245] Train net output #54: loss2/accuracy05 = 0.5
I0607 05:42:29.587684 32403 solver.cpp:245] Train net output #55: loss2/accuracy06 = 0.75
I0607 05:42:29.587697 32403 solver.cpp:245] Train net output #56: loss2/accuracy07 = 0.75
I0607 05:42:29.587708 32403 solver.cpp:245] Train net output #57: loss2/accuracy08 = 0.875
I0607 05:42:29.587720 32403 solver.cpp:245] Train net output #58: loss2/accuracy09 = 0.875
I0607 05:42:29.587733 32403 solver.cpp:245] Train net output #59: loss2/accuracy10 = 1
I0607 05:42:29.587744 32403 solver.cpp:245] Train net output #60: loss2/accuracy11 = 1
I0607 05:42:29.587756 32403 solver.cpp:245] Train net output #61: loss2/accuracy12 = 1
I0607 05:42:29.587767 32403 solver.cpp:245] Train net output #62: loss2/accuracy13 = 1
I0607 05:42:29.587779 32403 solver.cpp:245] Train net output #63: loss2/accuracy14 = 1
I0607 05:42:29.587791 32403 solver.cpp:245] Train net output #64: loss2/accuracy15 = 1
I0607 05:42:29.587803 32403 solver.cpp:245] Train net output #65: loss2/accuracy16 = 1
I0607 05:42:29.587815 32403 solver.cpp:245] Train net output #66: loss2/accuracy17 = 1
I0607 05:42:29.587827 32403 solver.cpp:245] Train net output #67: loss2/accuracy18 = 1
I0607 05:42:29.587839 32403 solver.cpp:245] Train net output #68: loss2/accuracy19 = 1
I0607 05:42:29.587852 32403 solver.cpp:245] Train net output #69: loss2/accuracy20 = 1
I0607 05:42:29.587863 32403 solver.cpp:245] Train net output #70: loss2/accuracy21 = 1
I0607 05:42:29.587878 32403 solver.cpp:245] Train net output #71: loss2/accuracy22 = 1
I0607 05:42:29.587891 32403 solver.cpp:245] Train net output #72: loss2/accuracy_incl_empty = 0.869318
I0607 05:42:29.587903 32403 solver.cpp:245] Train net output #73: loss2/accuracy_top3 = 0.822222
I0607 05:42:29.587918 32403 solver.cpp:245] Train net output #74: loss2/cross_entropy_loss = 1.40842 (* 0.3 = 0.422525 loss)
I0607 05:42:29.587932 32403 solver.cpp:245] Train net output #75: loss2/cross_entropy_loss_incl_empty = 0.508333 (* 0.3 = 0.1525 loss)
I0607 05:42:29.587946 32403 solver.cpp:245] Train net output #76: loss2/loss01 = 1.21124 (* 0.0272727 = 0.0330338 loss)
I0607 05:42:29.587960 32403 solver.cpp:245] Train net output #77: loss2/loss02 = 1.51845 (* 0.0272727 = 0.0414124 loss)
I0607 05:42:29.587986 32403 solver.cpp:245] Train net output #78: loss2/loss03 = 1.54151 (* 0.0272727 = 0.0420412 loss)
I0607 05:42:29.588002 32403 solver.cpp:245] Train net output #79: loss2/loss04 = 1.73028 (* 0.0272727 = 0.0471894 loss)
I0607 05:42:29.588016 32403 solver.cpp:245] Train net output #80: loss2/loss05 = 2.2889 (* 0.0272727 = 0.0624245 loss)
I0607 05:42:29.588030 32403 solver.cpp:245] Train net output #81: loss2/loss06 = 1.06506 (* 0.0272727 = 0.0290472 loss)
I0607 05:42:29.588044 32403 solver.cpp:245] Train net output #82: loss2/loss07 = 1.11499 (* 0.0272727 = 0.0304089 loss)
I0607 05:42:29.588058 32403 solver.cpp:245] Train net output #83: loss2/loss08 = 0.781693 (* 0.0272727 = 0.0213189 loss)
I0607 05:42:29.588073 32403 solver.cpp:245] Train net output #84: loss2/loss09 = 0.417263 (* 0.0272727 = 0.0113799 loss)
I0607 05:42:29.588086 32403 solver.cpp:245] Train net output #85: loss2/loss10 = 0.210767 (* 0.0272727 = 0.0057482 loss)
I0607 05:42:29.588100 32403 solver.cpp:245] Train net output #86: loss2/loss11 = 0.095405 (* 0.0272727 = 0.00260195 loss)
I0607 05:42:29.588114 32403 solver.cpp:245] Train net output
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment