Notes

import random

# training loop

# For each epoch...
for epoch_i in range(0, epochs):
# Perform one full pass over the training set.

print("")
print('======== Epoch {:} / {:} ========'.format(epoch_i + 1, epochs))
print('Training...')

# Reset the total loss for this epoch.
total_train_loss = 0

# Put the model into training mode.
model.train()

# For each batch of training data...
num_batches = int(len(train_set)/batch_size)
if int(len(train_set)/batch_size)!=len(train_set)/batch_size:
num_batches+=1

for i in range(num_batches):
end_index = min(batch_size * (i+1), len(train_set))
batch = train_set[i*batch_size:end_index]
input_id_tensors = torch.stack([data[0] for data in batch])
input_mask_tensors = torch.stack([data[1] for data in batch])
label_tensors = torch.stack([data[2] for data in batch])

# Move tensors to the GPU
b_input_ids = input_id_tensors.to(device)
b_input_mask = input_mask_tensors.to(device)
b_labels = label_tensors.to(device)

# Clear the previously calculated gradient
model.zero_grad()

# Perform a forward pass (evaluate the model on this training batch).
outputs = model(b_input_ids,
token_type_ids=None,
attention_mask=b_input_mask,
labels=b_labels)
loss = outputs.loss
logits = outputs.logits

total_train_loss += loss.item()

# Perform a backward pass to calculate the gradients.
loss.backward()

# Update parameters and take a step using the computed gradient.
optimizer.step()

# ========================================
# Validation
# ========================================
# After the completion of each training epoch, measure our performance on
# our validation set. Implement this function in the cell above.
print(f"Total loss: {total_train_loss}")
val_acc = get_validation_performance(val_set)
print(f"Validation accuracy: {val_acc}")

print("")
print("Training complete!")