You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem? Please describe.
When training on large datasets, it's difficult to know how far along training is and whether or not I should stop training to tweak parameters to speed up the training process.
Describe the solution you'd like
Add tqdm progress bar to training progress. See context section. I have a suggested patch to trainer.py that adds this. Happy to PR it in, or feel free to take as you see appropriate.
Describe alternatives you've considered
One alternative is a manual print. But tqdm is the standard for this typically.
I attached a patch file that adds an optional show_progress argument in a similar manner to what's used in DataLoader already. Note: This patch only shows it for the Trainer class, not the child classes. That is because this is for an example only.
The main change is this:
- for batch_idx, interaction in enumerate(train_data):+ iter_data = (+ tqdm(+ enumerate(train_data),+ total=len(train_data),+ desc=f"Train {epoch_idx:>5}",+ )+ if show_progress+ else enumerate(train_data)+ )+ for batch_idx, interaction in iter_data:
Here is an example output while training a larger dataset. Scroll right to see full output.
Thanks for your insightful commment. Using tqdm is a good way to add this feature. We'd appreciate it if you would like to PR it in branch 0.1.x. By the way, any PRs are welcome.
Note: trainer.py has been refactor in #521 , be careful to add this feature.
Is your feature request related to a problem? Please describe.
When training on large datasets, it's difficult to know how far along training is and whether or not I should stop training to tweak parameters to speed up the training process.
Describe the solution you'd like
Add tqdm progress bar to training progress. See context section. I have a suggested patch to
trainer.py
that adds this. Happy to PR it in, or feel free to take as you see appropriate.Describe alternatives you've considered
One alternative is a manual print. But tqdm is the standard for this typically.
Additional context
trainer.py.patch.txt
I attached a patch file that adds an optional
show_progress
argument in a similar manner to what's used in DataLoader already. Note: This patch only shows it for theTrainer
class, not the child classes. That is because this is for an example only.The main change is this:
Here is an example output while training a larger dataset. Scroll right to see full output.
The text was updated successfully, but these errors were encountered: