Add learning rate config support for gpt-oss models #197
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
bledden
force-pushed
the
fix/gpt-oss-lr-configs
branch
from
9206be1 to
52f8c0d
Compare
The get_lr() function was raising an AssertionError for openai/gpt-oss-20b and openai/gpt-oss-120b models because they weren't included in the model name checks. This adds: - Hidden size lookup for gpt-oss models (both use 2880) - Learning rate scaling exponent for gpt-oss (using 0.0775, same as Qwen since both are MoE architectures) Fixes thinking-machines-lab/tinker-feedback#49
bledden
force-pushed
the
fix/gpt-oss-lr-configs
branch
from
52f8c0d to
118fd43
Compare
Use exact model name matching instead of partial string matching to prevent unintended matches with future model variants. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.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.
Summary
The
get_lr()function inhyperparam_utils.pywas missing support for the gpt-oss model family, causing it to fail with anAssertionErrorwhen users tried to get learning rate recommendations foropenai/gpt-oss-20boropenai/gpt-oss-120b.This PR adds the missing configuration:
Test plan
get_lr("openai/gpt-oss-20b")returns a valid learning rate instead of raising an errorget_lr("openai/gpt-oss-120b")works correctlyFixes thinking-machines-lab/tinker-feedback#49