GRPO loss #454
Merged
GRPO loss #454
Conversation
This file contains hidden or 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
…ow/Fast-LLM into train_only_layer_losses
…ow/Fast-LLM into train_only_layer_losses
Removed the targets, class, moved tragets processing to losses, made loss masks more explicit
rafapi
reviewed
Jan 20, 2026
|
|
||
| # Policy loss | ||
| # TODO: advantages or rewards? | ||
| log_ratio_old = torch.exp(target_logprobs - old_logprobs) |
Collaborator
There was a problem hiding this comment.
this would read better as ratio_new_old = torch.exp(target_logprobs - old_logprobs)
rafapi
reviewed
Jan 20, 2026
| target_logprobs = torch.gather(logprobs, dim=2, index=labels.unsqueeze(2)).squeeze(2) | ||
|
|
||
| # Policy loss | ||
| # TODO: advantages or rewards? |
rafapi
reviewed
Jan 20, 2026
| ) | ||
|
|
||
| # TODO: tokens_weights = 1/batch_size ? | ||
| # TODO: Reduce loss? |
Collaborator
There was a problem hiding this comment.
need to sum over tokens and apply mask
rafapi
reviewed
Jan 20, 2026
| torch.clamp(log_ratio_old, 1 - self.epsilon_low, 1 + self.epsilon_high) * advantage, | ||
| ) | ||
|
|
||
| # TODO: tokens_weights = 1/batch_size ? |
Collaborator
There was a problem hiding this comment.
i think so, we do that for the simple case
rafapi
reviewed
Jan 20, 2026
|
|
||
| # TODO: tokens_weights = 1/batch_size ? | ||
| # TODO: Reduce loss? | ||
| loss = loss / batch_size # 1 x (BxL) x 1 |
Collaborator
There was a problem hiding this comment.
loss = -loss - we want to maximise the objective
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Outdated suggestions cannot be applied.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.
✨ Description