Fix OLMoE routing and Mistral4 RoPE dimensions#45366
Closed
owwll wants to merge 1 commit intohuggingface:mainfrom
Closed
Fix OLMoE routing and Mistral4 RoPE dimensions#45366owwll wants to merge 1 commit intohuggingface:mainfrom
owwll wants to merge 1 commit intohuggingface:mainfrom
Conversation
Contributor
|
[For maintainers] Suggested jobs to run (before merge) run-slow: mistral4, olmoe |
Contributor
|
View the CircleCI Test Summary for this PR: https://huggingface.co/spaces/transformers-community/circle-ci-viz?pr=45366&sha=de9d2f |
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This PR addresses two separate issues:
Fixes a bug in
Mistral4RoPE dimension calculation.The
Mistral4RotaryEmbeddingwas incorrectly using the fullhead_dimto calculate the rotary dimension, instead of respecting thepartial_rotary_factor. This has been corrected to useint(head_dim * config.rope_parameters.get("partial_rotary_factor", 1.0)). A new regression test has been added totest_modeling_mistral4.pyto cover this case.Fixes bugs in the
OLMoEmodel.OlmoeAttentionlayer was incorrectly initializing its normalization layers (q_normandk_norm) with the wrong dimension. This has been fixed.OlmoeTopKRouterwas returning softmax probabilities instead of raw logits, which is inconsistent with other router implementations. This has been corrected.test_modeling_olmoe.pyto validate these fixes.This contribution follows the project's contribution guidelines.