gistfile1.txt

Can not use cuDNN on context None: Disabled by dnn.enabled flag
Mapped name None to device cuda0: GeForce GTX TITAN X (0000:83:00.0)
11/14/2018 02:45:18 PM In order to work for big datasets fix https://github.com/Theano/Theano/pull/5721 should be applied to theano.
11/14/2018 02:45:18 PM loading data from dumped file...
11/14/2018 02:45:18 PM loading data finished!
11/14/2018 02:45:18 PM stacking training, dev and test features and creating indices...
11/14/2018 02:45:18 PM running mlp with graph conv...
11/14/2018 02:45:18 PM highway is True
11/14/2018 02:45:18 PM Graphconv model input size 9467, output size 32 and hidden layers [300, 300, 300] regul 0.0 dropout 0.5.
11/14/2018 02:45:18 PM 3 gconv layers
/home/felixgwu/.conda/envs/geo/lib/python2.7/site-packages/lasagne/layers/helper.py:216: UserWarning: get_output() was called with unused kwargs:
	A
  % "\n\t".join(suggestions))
/share/felixgwu/projects/gcn_project/geographconv/gcnmodel.py:402: UserWarning: theano.function was asked to create a function computing outputs given certain inputs, but the provided input variable at index 1 is not part of the computational graph needed to compute the outputs: SparseVariable{csr,float32}.
To make this warning into an error, you can pass the parameter on_unused_input='raise' to theano.function. To disable it completely, use on_unused_input='ignore'.
  self.f_gates.append(theano.function([self.X_sym, self.A_sym], self.gate_outputs[i], on_unused_input='warn'))
11/14/2018 02:45:29 PM ***********percentile 1.000000 ******************
11/14/2018 02:45:29 PM 5685 training samples
11/14/2018 02:45:29 PM training for 10000 epochs with batch size 500
/home/felixgwu/.conda/envs/geo/lib/python2.7/site-packages/theano/tensor/subtensor.py:2339: FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated; use `arr[tuple(seq)]` instead of `arr[seq]`. In the future this will be interpreted as an array index, `arr[np.array(seq)]`, which will result either in an error or a different result.
  out[0][inputs[2:]] = inputs[1]
11/14/2018 02:46:22 PM validation results went down. early stopping ...
11/14/2018 02:46:22 PM dev results:
11/14/2018 02:46:22 PM Mean: 541 Median: 104 Acc@161: 54
11/14/2018 02:46:22 PM test results:
11/14/2018 02:46:22 PM Mean: 550 Median: 96 Acc@161: 55
Can not use cuDNN on context None: Disabled by dnn.enabled flag
Mapped name None to device cuda0: GeForce GTX TITAN X (0000:83:00.0)
11/14/2018 02:46:28 PM In order to work for big datasets fix https://github.com/Theano/Theano/pull/5721 should be applied to theano.
11/14/2018 02:46:28 PM loading data from dumped file...
11/14/2018 02:46:28 PM loading data finished!
11/14/2018 02:46:28 PM stacking training, dev and test features and creating indices...
11/14/2018 02:46:29 PM running mlp with graph conv...
11/14/2018 02:46:29 PM highway is True
11/14/2018 02:46:29 PM Graphconv model input size 9467, output size 32 and hidden layers [300, 300, 300] regul 0.0 dropout 0.5.
11/14/2018 02:46:29 PM 3 gconv layers
/home/felixgwu/.conda/envs/geo/lib/python2.7/site-packages/lasagne/layers/helper.py:216: UserWarning: get_output() was called with unused kwargs:
	A
  % "\n\t".join(suggestions))
/share/felixgwu/projects/gcn_project/geographconv/gcnmodel.py:402: UserWarning: theano.function was asked to create a function computing outputs given certain inputs, but the provided input variable at index 1 is not part of the computational graph needed to compute the outputs: SparseVariable{csr,float32}.
To make this warning into an error, you can pass the parameter on_unused_input='raise' to theano.function. To disable it completely, use on_unused_input='ignore'.
  self.f_gates.append(theano.function([self.X_sym, self.A_sym], self.gate_outputs[i], on_unused_input='warn'))
11/14/2018 02:46:40 PM ***********percentile 1.000000 ******************
11/14/2018 02:46:40 PM 5685 training samples
11/14/2018 02:46:40 PM training for 10000 epochs with batch size 500
/home/felixgwu/.conda/envs/geo/lib/python2.7/site-packages/theano/tensor/subtensor.py:2339: FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated; use `arr[tuple(seq)]` instead of `arr[seq]`. In the future this will be interpreted as an array index, `arr[np.array(seq)]`, which will result either in an error or a different result.
  out[0][inputs[2:]] = inputs[1]
11/14/2018 02:47:30 PM validation results went down. early stopping ...
11/14/2018 02:47:30 PM dev results:
11/14/2018 02:47:31 PM Mean: 541 Median: 102 Acc@161: 55
11/14/2018 02:47:31 PM test results:
11/14/2018 02:47:31 PM Mean: 565 Median: 98 Acc@161: 54
Can not use cuDNN on context None: Disabled by dnn.enabled flag
Mapped name None to device cuda0: GeForce GTX TITAN X (0000:83:00.0)
11/14/2018 02:47:37 PM In order to work for big datasets fix https://github.com/Theano/Theano/pull/5721 should be applied to theano.
11/14/2018 02:47:37 PM loading data from dumped file...
11/14/2018 02:47:38 PM loading data finished!
11/14/2018 02:47:38 PM stacking training, dev and test features and creating indices...
11/14/2018 02:47:38 PM running mlp with graph conv...
11/14/2018 02:47:38 PM highway is True
11/14/2018 02:47:38 PM Graphconv model input size 9467, output size 32 and hidden layers [300, 300, 300] regul 0.0 dropout 0.5.
11/14/2018 02:47:38 PM 3 gconv layers
/home/felixgwu/.conda/envs/geo/lib/python2.7/site-packages/lasagne/layers/helper.py:216: UserWarning: get_output() was called with unused kwargs:
	A
  % "\n\t".join(suggestions))
/share/felixgwu/projects/gcn_project/geographconv/gcnmodel.py:402: UserWarning: theano.function was asked to create a function computing outputs given certain inputs, but the provided input variable at index 1 is not part of the computational graph needed to compute the outputs: SparseVariable{csr,float32}.
To make this warning into an error, you can pass the parameter on_unused_input='raise' to theano.function. To disable it completely, use on_unused_input='ignore'.
  self.f_gates.append(theano.function([self.X_sym, self.A_sym], self.gate_outputs[i], on_unused_input='warn'))
11/14/2018 02:47:49 PM ***********percentile 1.000000 ******************
11/14/2018 02:47:49 PM 5685 training samples
11/14/2018 02:47:49 PM training for 10000 epochs with batch size 500
/home/felixgwu/.conda/envs/geo/lib/python2.7/site-packages/theano/tensor/subtensor.py:2339: FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated; use `arr[tuple(seq)]` instead of `arr[seq]`. In the future this will be interpreted as an array index, `arr[np.array(seq)]`, which will result either in an error or a different result.
  out[0][inputs[2:]] = inputs[1]
11/14/2018 02:48:34 PM validation results went down. early stopping ...
11/14/2018 02:48:34 PM dev results:
11/14/2018 02:48:35 PM Mean: 529 Median: 102 Acc@161: 55
11/14/2018 02:48:35 PM test results:
11/14/2018 02:48:35 PM Mean: 549 Median: 94 Acc@161: 55
Can not use cuDNN on context None: Disabled by dnn.enabled flag
Mapped name None to device cuda0: GeForce GTX TITAN X (0000:83:00.0)
11/14/2018 02:48:40 PM In order to work for big datasets fix https://github.com/Theano/Theano/pull/5721 should be applied to theano.
11/14/2018 02:48:40 PM loading data from dumped file...
11/14/2018 02:48:41 PM loading data finished!
11/14/2018 02:48:41 PM stacking training, dev and test features and creating indices...
11/14/2018 02:48:41 PM running mlp with graph conv...
11/14/2018 02:48:41 PM highway is True
11/14/2018 02:48:41 PM Graphconv model input size 9467, output size 32 and hidden layers [300, 300, 300] regul 0.0 dropout 0.5.
11/14/2018 02:48:41 PM 3 gconv layers
/home/felixgwu/.conda/envs/geo/lib/python2.7/site-packages/lasagne/layers/helper.py:216: UserWarning: get_output() was called with unused kwargs:
	A
  % "\n\t".join(suggestions))
/share/felixgwu/projects/gcn_project/geographconv/gcnmodel.py:402: UserWarning: theano.function was asked to create a function computing outputs given certain inputs, but the provided input variable at index 1 is not part of the computational graph needed to compute the outputs: SparseVariable{csr,float32}.
To make this warning into an error, you can pass the parameter on_unused_input='raise' to theano.function. To disable it completely, use on_unused_input='ignore'.
  self.f_gates.append(theano.function([self.X_sym, self.A_sym], self.gate_outputs[i], on_unused_input='warn'))
11/14/2018 02:48:52 PM ***********percentile 1.000000 ******************
11/14/2018 02:48:52 PM 5685 training samples
11/14/2018 02:48:52 PM training for 10000 epochs with batch size 500
/home/felixgwu/.conda/envs/geo/lib/python2.7/site-packages/theano/tensor/subtensor.py:2339: FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated; use `arr[tuple(seq)]` instead of `arr[seq]`. In the future this will be interpreted as an array index, `arr[np.array(seq)]`, which will result either in an error or a different result.
  out[0][inputs[2:]] = inputs[1]
11/14/2018 02:49:41 PM validation results went down. early stopping ...
11/14/2018 02:49:41 PM dev results:
11/14/2018 02:49:41 PM Mean: 534 Median: 102 Acc@161: 55
11/14/2018 02:49:41 PM test results:
11/14/2018 02:49:41 PM Mean: 541 Median: 95 Acc@161: 55
Can not use cuDNN on context None: Disabled by dnn.enabled flag
Mapped name None to device cuda0: GeForce GTX TITAN X (0000:83:00.0)
11/14/2018 02:49:47 PM In order to work for big datasets fix https://github.com/Theano/Theano/pull/5721 should be applied to theano.
11/14/2018 02:49:47 PM loading data from dumped file...
11/14/2018 02:49:47 PM loading data finished!
11/14/2018 02:49:47 PM stacking training, dev and test features and creating indices...
11/14/2018 02:49:47 PM running mlp with graph conv...
11/14/2018 02:49:47 PM highway is True
11/14/2018 02:49:47 PM Graphconv model input size 9467, output size 32 and hidden layers [300, 300, 300] regul 0.0 dropout 0.5.
11/14/2018 02:49:47 PM 3 gconv layers
/home/felixgwu/.conda/envs/geo/lib/python2.7/site-packages/lasagne/layers/helper.py:216: UserWarning: get_output() was called with unused kwargs:
	A
  % "\n\t".join(suggestions))
/share/felixgwu/projects/gcn_project/geographconv/gcnmodel.py:402: UserWarning: theano.function was asked to create a function computing outputs given certain inputs, but the provided input variable at index 1 is not part of the computational graph needed to compute the outputs: SparseVariable{csr,float32}.
To make this warning into an error, you can pass the parameter on_unused_input='raise' to theano.function. To disable it completely, use on_unused_input='ignore'.
  self.f_gates.append(theano.function([self.X_sym, self.A_sym], self.gate_outputs[i], on_unused_input='warn'))
11/14/2018 02:50:00 PM ***********percentile 1.000000 ******************
11/14/2018 02:50:00 PM 5685 training samples
11/14/2018 02:50:00 PM training for 10000 epochs with batch size 500
/home/felixgwu/.conda/envs/geo/lib/python2.7/site-packages/theano/tensor/subtensor.py:2339: FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated; use `arr[tuple(seq)]` instead of `arr[seq]`. In the future this will be interpreted as an array index, `arr[np.array(seq)]`, which will result either in an error or a different result.
  out[0][inputs[2:]] = inputs[1]
11/14/2018 02:50:49 PM validation results went down. early stopping ...
11/14/2018 02:50:49 PM dev results:
11/14/2018 02:50:49 PM Mean: 530 Median: 102 Acc@161: 54
11/14/2018 02:50:49 PM test results:
11/14/2018 02:50:49 PM Mean: 555 Median: 97 Acc@161: 54
Can not use cuDNN on context None: Disabled by dnn.enabled flag
Mapped name None to device cuda0: GeForce GTX TITAN X (0000:83:00.0)
11/14/2018 02:50:55 PM In order to work for big datasets fix https://github.com/Theano/Theano/pull/5721 should be applied to theano.
11/14/2018 02:50:55 PM loading data from dumped file...
11/14/2018 02:50:55 PM loading data finished!
11/14/2018 02:50:55 PM stacking training, dev and test features and creating indices...
11/14/2018 02:50:55 PM running mlp with graph conv...
11/14/2018 02:50:55 PM highway is True
11/14/2018 02:50:55 PM Graphconv model input size 9467, output size 32 and hidden layers [300, 300, 300] regul 0.0 dropout 0.5.
11/14/2018 02:50:55 PM 3 gconv layers
/home/felixgwu/.conda/envs/geo/lib/python2.7/site-packages/lasagne/layers/helper.py:216: UserWarning: get_output() was called with unused kwargs:
	A
  % "\n\t".join(suggestions))
/share/felixgwu/projects/gcn_project/geographconv/gcnmodel.py:402: UserWarning: theano.function was asked to create a function computing outputs given certain inputs, but the provided input variable at index 1 is not part of the computational graph needed to compute the outputs: SparseVariable{csr,float32}.
To make this warning into an error, you can pass the parameter on_unused_input='raise' to theano.function. To disable it completely, use on_unused_input='ignore'.
  self.f_gates.append(theano.function([self.X_sym, self.A_sym], self.gate_outputs[i], on_unused_input='warn'))
11/14/2018 02:51:06 PM ***********percentile 1.000000 ******************
11/14/2018 02:51:06 PM 5685 training samples
11/14/2018 02:51:06 PM training for 10000 epochs with batch size 500
/home/felixgwu/.conda/envs/geo/lib/python2.7/site-packages/theano/tensor/subtensor.py:2339: FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated; use `arr[tuple(seq)]` instead of `arr[seq]`. In the future this will be interpreted as an array index, `arr[np.array(seq)]`, which will result either in an error or a different result.
  out[0][inputs[2:]] = inputs[1]
11/14/2018 02:51:56 PM validation results went down. early stopping ...
11/14/2018 02:51:56 PM dev results:
11/14/2018 02:51:56 PM Mean: 554 Median: 100 Acc@161: 55
11/14/2018 02:51:56 PM test results:
11/14/2018 02:51:57 PM Mean: 567 Median: 96 Acc@161: 55
Can not use cuDNN on context None: Disabled by dnn.enabled flag
Mapped name None to device cuda0: GeForce GTX TITAN X (0000:83:00.0)
11/14/2018 02:52:05 PM In order to work for big datasets fix https://github.com/Theano/Theano/pull/5721 should be applied to theano.
11/14/2018 02:52:05 PM loading data from dumped file...
11/14/2018 02:52:06 PM loading data finished!
11/14/2018 02:52:06 PM stacking training, dev and test features and creating indices...
11/14/2018 02:52:06 PM running mlp with graph conv...
11/14/2018 02:52:06 PM highway is True
11/14/2018 02:52:06 PM Graphconv model input size 9467, output size 32 and hidden layers [300, 300, 300] regul 0.0 dropout 0.5.
11/14/2018 02:52:06 PM 3 gconv layers
/home/felixgwu/.conda/envs/geo/lib/python2.7/site-packages/lasagne/layers/helper.py:216: UserWarning: get_output() was called with unused kwargs:
	A
  % "\n\t".join(suggestions))
/share/felixgwu/projects/gcn_project/geographconv/gcnmodel.py:402: UserWarning: theano.function was asked to create a function computing outputs given certain inputs, but the provided input variable at index 1 is not part of the computational graph needed to compute the outputs: SparseVariable{csr,float32}.
To make this warning into an error, you can pass the parameter on_unused_input='raise' to theano.function. To disable it completely, use on_unused_input='ignore'.
  self.f_gates.append(theano.function([self.X_sym, self.A_sym], self.gate_outputs[i], on_unused_input='warn'))
11/14/2018 02:52:21 PM ***********percentile 1.000000 ******************
11/14/2018 02:52:21 PM 5685 training samples
11/14/2018 02:52:21 PM training for 10000 epochs with batch size 500
/home/felixgwu/.conda/envs/geo/lib/python2.7/site-packages/theano/tensor/subtensor.py:2339: FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated; use `arr[tuple(seq)]` instead of `arr[seq]`. In the future this will be interpreted as an array index, `arr[np.array(seq)]`, which will result either in an error or a different result.
  out[0][inputs[2:]] = inputs[1]
11/14/2018 02:53:11 PM validation results went down. early stopping ...
11/14/2018 02:53:11 PM dev results:
11/14/2018 02:53:11 PM Mean: 536 Median: 102 Acc@161: 55
11/14/2018 02:53:11 PM test results:
11/14/2018 02:53:11 PM Mean: 552 Median: 97 Acc@161: 55
Can not use cuDNN on context None: Disabled by dnn.enabled flag
Mapped name None to device cuda0: GeForce GTX TITAN X (0000:83:00.0)
11/14/2018 02:53:17 PM In order to work for big datasets fix https://github.com/Theano/Theano/pull/5721 should be applied to theano.
11/14/2018 02:53:17 PM loading data from dumped file...
11/14/2018 02:53:17 PM loading data finished!
11/14/2018 02:53:17 PM stacking training, dev and test features and creating indices...
11/14/2018 02:53:17 PM running mlp with graph conv...
11/14/2018 02:53:17 PM highway is True
11/14/2018 02:53:17 PM Graphconv model input size 9467, output size 32 and hidden layers [300, 300, 300] regul 0.0 dropout 0.5.
11/14/2018 02:53:17 PM 3 gconv layers
/home/felixgwu/.conda/envs/geo/lib/python2.7/site-packages/lasagne/layers/helper.py:216: UserWarning: get_output() was called with unused kwargs:
	A
  % "\n\t".join(suggestions))
/share/felixgwu/projects/gcn_project/geographconv/gcnmodel.py:402: UserWarning: theano.function was asked to create a function computing outputs given certain inputs, but the provided input variable at index 1 is not part of the computational graph needed to compute the outputs: SparseVariable{csr,float32}.
To make this warning into an error, you can pass the parameter on_unused_input='raise' to theano.function. To disable it completely, use on_unused_input='ignore'.
  self.f_gates.append(theano.function([self.X_sym, self.A_sym], self.gate_outputs[i], on_unused_input='warn'))
11/14/2018 02:53:28 PM ***********percentile 1.000000 ******************
11/14/2018 02:53:28 PM 5685 training samples
11/14/2018 02:53:28 PM training for 10000 epochs with batch size 500
/home/felixgwu/.conda/envs/geo/lib/python2.7/site-packages/theano/tensor/subtensor.py:2339: FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated; use `arr[tuple(seq)]` instead of `arr[seq]`. In the future this will be interpreted as an array index, `arr[np.array(seq)]`, which will result either in an error or a different result.
  out[0][inputs[2:]] = inputs[1]
11/14/2018 02:54:17 PM validation results went down. early stopping ...
11/14/2018 02:54:17 PM dev results:
11/14/2018 02:54:17 PM Mean: 537 Median: 101 Acc@161: 55
11/14/2018 02:54:17 PM test results:
11/14/2018 02:54:17 PM Mean: 551 Median: 96 Acc@161: 55
Can not use cuDNN on context None: Disabled by dnn.enabled flag
Mapped name None to device cuda0: GeForce GTX TITAN X (0000:83:00.0)
11/14/2018 02:54:22 PM In order to work for big datasets fix https://github.com/Theano/Theano/pull/5721 should be applied to theano.
11/14/2018 02:54:22 PM loading data from dumped file...
11/14/2018 02:54:23 PM loading data finished!
11/14/2018 02:54:23 PM stacking training, dev and test features and creating indices...
11/14/2018 02:54:23 PM running mlp with graph conv...
11/14/2018 02:54:23 PM highway is True
11/14/2018 02:54:23 PM Graphconv model input size 9467, output size 32 and hidden layers [300, 300, 300] regul 0.0 dropout 0.5.
11/14/2018 02:54:23 PM 3 gconv layers
/home/felixgwu/.conda/envs/geo/lib/python2.7/site-packages/lasagne/layers/helper.py:216: UserWarning: get_output() was called with unused kwargs:
	A
  % "\n\t".join(suggestions))
/share/felixgwu/projects/gcn_project/geographconv/gcnmodel.py:402: UserWarning: theano.function was asked to create a function computing outputs given certain inputs, but the provided input variable at index 1 is not part of the computational graph needed to compute the outputs: SparseVariable{csr,float32}.
To make this warning into an error, you can pass the parameter on_unused_input='raise' to theano.function. To disable it completely, use on_unused_input='ignore'.
  self.f_gates.append(theano.function([self.X_sym, self.A_sym], self.gate_outputs[i], on_unused_input='warn'))
11/14/2018 02:54:35 PM ***********percentile 1.000000 ******************
11/14/2018 02:54:35 PM 5685 training samples
11/14/2018 02:54:35 PM training for 10000 epochs with batch size 500
/home/felixgwu/.conda/envs/geo/lib/python2.7/site-packages/theano/tensor/subtensor.py:2339: FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated; use `arr[tuple(seq)]` instead of `arr[seq]`. In the future this will be interpreted as an array index, `arr[np.array(seq)]`, which will result either in an error or a different result.
  out[0][inputs[2:]] = inputs[1]
11/14/2018 02:55:20 PM validation results went down. early stopping ...
11/14/2018 02:55:20 PM dev results:
11/14/2018 02:55:20 PM Mean: 534 Median: 100 Acc@161: 55
11/14/2018 02:55:20 PM test results:
11/14/2018 02:55:20 PM Mean: 554 Median: 96 Acc@161: 54
Can not use cuDNN on context None: Disabled by dnn.enabled flag
Mapped name None to device cuda0: GeForce GTX TITAN X (0000:83:00.0)
11/14/2018 02:55:25 PM In order to work for big datasets fix https://github.com/Theano/Theano/pull/5721 should be applied to theano.
11/14/2018 02:55:25 PM loading data from dumped file...
11/14/2018 02:55:26 PM loading data finished!
11/14/2018 02:55:26 PM stacking training, dev and test features and creating indices...
11/14/2018 02:55:26 PM running mlp with graph conv...
11/14/2018 02:55:26 PM highway is True
11/14/2018 02:55:26 PM Graphconv model input size 9467, output size 32 and hidden layers [300, 300, 300] regul 0.0 dropout 0.5.
11/14/2018 02:55:26 PM 3 gconv layers
/home/felixgwu/.conda/envs/geo/lib/python2.7/site-packages/lasagne/layers/helper.py:216: UserWarning: get_output() was called with unused kwargs:
	A
  % "\n\t".join(suggestions))
/share/felixgwu/projects/gcn_project/geographconv/gcnmodel.py:402: UserWarning: theano.function was asked to create a function computing outputs given certain inputs, but the provided input variable at index 1 is not part of the computational graph needed to compute the outputs: SparseVariable{csr,float32}.
To make this warning into an error, you can pass the parameter on_unused_input='raise' to theano.function. To disable it completely, use on_unused_input='ignore'.
  self.f_gates.append(theano.function([self.X_sym, self.A_sym], self.gate_outputs[i], on_unused_input='warn'))
11/14/2018 02:55:36 PM ***********percentile 1.000000 ******************
11/14/2018 02:55:36 PM 5685 training samples
11/14/2018 02:55:36 PM training for 10000 epochs with batch size 500
/home/felixgwu/.conda/envs/geo/lib/python2.7/site-packages/theano/tensor/subtensor.py:2339: FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated; use `arr[tuple(seq)]` instead of `arr[seq]`. In the future this will be interpreted as an array index, `arr[np.array(seq)]`, which will result either in an error or a different result.
  out[0][inputs[2:]] = inputs[1]
11/14/2018 02:56:22 PM validation results went down. early stopping ...
11/14/2018 02:56:22 PM dev results:
11/14/2018 02:56:22 PM Mean: 555 Median: 100 Acc@161: 55
11/14/2018 02:56:22 PM test results:
11/14/2018 02:56:22 PM Mean: 562 Median: 98 Acc@161: 54