Log Token distribution of Query / Response #295
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
The documentation is not available anymore as the PR was closed or merged. |
lvwerra
reviewed
Apr 12, 2023
There was a problem hiding this comment.
Thanks for adding this @natolambert! Generally looks good, added two comments.
trl/trainer/ppo_config.py
Outdated
| @@ -73,6 +73,10 @@ class PPOConfig(object): | |||
| "help": "Log with either 'wandb' or 'tensorboard', check https://huggingface.co/docs/accelerate/usage_guides/tracking for more details" | |||
| }, | |||
| ) | |||
| log_token_distribution: Optional[bool] = field( | |||
There was a problem hiding this comment.
I would not add another config and just always log it. It's cheap to do.
trl/trainer/ppo_trainer.py
Outdated
| stats["tokens/query_len_mean"] = torch.mean(query_lens).cpu().numpy().item() | ||
| stats["tokens/query_len_std"] = torch.std(query_lens).cpu().numpy().item() | ||
| stats["tokens/responses_len_mean"] = torch.mean(response_lens).cpu().numpy().item() | ||
| stats["tokens/responses_len_std"] = torch.std(response_lens).cpu().numpy().item() |
There was a problem hiding this comment.
similar to rewards i would also log the distribution so we can look at histograms in the logs.
|
KK should be good, example here: https://wandb.ai/huggingface/trl/runs/r2w2101u?workspace=user-natolambert |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Here's an example run https://wandb.ai/huggingface/trl/runs/gjxlr3on?workspace=user-natolambert