mobile_gallery_pytorch_09_train.py

loader = {'train': train_loader, 'val': val_loader}

epochs = 5
log_interval = 2

# Let's train the model for 5 epochs !
train_losses, val_losses, batch_train_losses, batch_val_losses = trainer(loader, model, loss_fn, optimizer, epochs = epochs, log_interval = log_interval)

# Ploting the epoch losses
plt.plot(train_losses)
plt.plot(val_losses)
plt.legend(['train losses', 'val_losses'])
plt.title('Loss vs Epoch')

plt.figure()
plt.plot(batch_train_losses)
plt.title('batch_train_losses')

plt.figure()
plt.plot(batch_val_losses)
plt.title('batch_val_losses')


# Saving the model(architecture and weights)
torch.save(model, 'stage1.pth')

mobile_gallery_pytorch_09_train_output.py

Training started...
epoch >>> 1/5
___TRAINING___
batch_loss at batch_idx 01/16: 1.8830947875976562
batch_loss at batch_idx 03/16: 1.7403203248977661
batch_loss at batch_idx 05/16: 1.7298262119293213
batch_loss at batch_idx 07/16: 1.5611944198608398
/opt/conda/lib/python3.6/site-packages/PIL/TiffImagePlugin.py:780: UserWarning: Corrupt EXIF data.  Expecting to read 4 bytes but only got 0. 
  warnings.warn(str(msg))
batch_loss at batch_idx 09/16: 1.5026812553405762
batch_loss at batch_idx 11/16: 1.3258850574493408
batch_loss at batch_idx 13/16: 1.2495317459106445
batch_loss at batch_idx 15/16: 1.305968165397644
>>> train loss at epoch 1/5: 1.5462412377120007
___VALIDATION___
batch_loss at batch_idx 01/16: 1.1894911527633667
batch_loss at batch_idx 03/16: 1.152886152267456
>>> val loss at epoch 1/5: 1.1689875689078504
=========================
epoch >>> 2/5
___TRAINING___
batch_loss at batch_idx 01/16: 1.0216645002365112
batch_loss at batch_idx 03/16: 1.0510553121566772
batch_loss at batch_idx 05/16: 0.9224186539649963
batch_loss at batch_idx 07/16: 0.8934431672096252
batch_loss at batch_idx 09/16: 0.8022943735122681
batch_loss at batch_idx 11/16: 0.8075667023658752
batch_loss at batch_idx 13/16: 0.7715064287185669
batch_loss at batch_idx 15/16: 0.8646692633628845
>>> train loss at epoch 2/5: 0.8703589484154471
___VALIDATION___
batch_loss at batch_idx 01/16: 0.6990357637405396
batch_loss at batch_idx 03/16: 0.6862280368804932
>>> val loss at epoch 2/5: 0.7019397386415737
=========================
epoch >>> 3/5
___TRAINING___
batch_loss at batch_idx 01/16: 0.6628543138504028
batch_loss at batch_idx 03/16: 0.5495110750198364
batch_loss at batch_idx 05/16: 0.4737720787525177
batch_loss at batch_idx 07/16: 0.5540937781333923
batch_loss at batch_idx 09/16: 0.5418666005134583
batch_loss at batch_idx 11/16: 0.483386754989624
batch_loss at batch_idx 13/16: 0.4547680914402008
batch_loss at batch_idx 15/16: 0.4926633834838867
>>> train loss at epoch 3/5: 0.5295033304116471
___VALIDATION___
batch_loss at batch_idx 01/16: 0.4764443635940552
batch_loss at batch_idx 03/16: 0.45040857791900635
>>> val loss at epoch 3/5: 0.4676538405455942
=========================
epoch >>> 4/5
___TRAINING___
batch_loss at batch_idx 01/16: 0.3784201443195343
batch_loss at batch_idx 03/16: 0.35650306940078735
batch_loss at batch_idx 05/16: 0.40147092938423157
batch_loss at batch_idx 07/16: 0.3184959292411804
batch_loss at batch_idx 09/16: 0.32096436619758606
batch_loss at batch_idx 11/16: 0.3386695683002472
batch_loss at batch_idx 13/16: 0.32678791880607605
batch_loss at batch_idx 15/16: 0.4038775861263275
>>> train loss at epoch 4/5: 0.3578496524703361
___VALIDATION___
batch_loss at batch_idx 01/16: 0.3420213460922241
batch_loss at batch_idx 03/16: 0.33064761757850647
>>> val loss at epoch 4/5: 0.3362177138722788
=========================
epoch >>> 5/5
___TRAINING___
batch_loss at batch_idx 01/16: 0.3416098356246948
batch_loss at batch_idx 03/16: 0.25589874386787415
batch_loss at batch_idx 05/16: 0.2259582132101059
batch_loss at batch_idx 07/16: 0.27338215708732605
batch_loss at batch_idx 09/16: 0.2513640224933624
batch_loss at batch_idx 11/16: 0.25437164306640625
batch_loss at batch_idx 13/16: 0.22256909310817719
batch_loss at batch_idx 15/16: 0.2105967402458191
>>> train loss at epoch 5/5: 0.2632993612836001
___VALIDATION___
batch_loss at batch_idx 01/16: 0.24774116277694702
batch_loss at batch_idx 03/16: 0.24713973701000214
>>> val loss at epoch 5/5: 0.258364795347837
=========================