Skip to content

made a change so that model can be saved when starting from checkpouint #128

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
mushahid-intesum wants to merge 3 commits into
base: main
Choose a base branch
from
Open

made a change so that model can be saved when starting from checkpouint #128

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
56f4f71
made a change so that model can be saved when starting from checkpouint
mushahid-intesum Oct 16, 2023
bffcf76
reduced reduntant code
mushahid-intesum Oct 20, 2023
2457c87
Merge branch 'coqui-ai:main' into checkpoint_restart_bug
mushahid-intesum Oct 20, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 15 additions & 0 deletions trainer/io.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -180,6 +180,20 @@ def save_best_model(
save_func=None,
**kwargs,
):
"""
Saves the best model based on the training losses

Compares the best loss to the current loss. If current loss is better than the previous loss, current is set to best loss

When starting from a saved checkpoint, the losses are stored in a dict like the following one
{train_loss: value, val_loss: value}

Needed to handle this when the model training is restarted from a checkpoint
"""

if isinstance(best_loss, dict):
best_loss = best_loss["train_loss"]

if current_loss < best_loss:
best_model_name = f"best_model_{current_step}.pth"
checkpoint_path = os.path.join(out_path, best_model_name)
Expand Down Expand Up @@ -208,6 +222,7 @@ def save_best_model(
shortcut_path = os.path.join(out_path, shortcut_name)
fs.copy(checkpoint_path, shortcut_path)
best_loss = current_loss

return best_loss


Expand Down