Cast tl.program_id(0) to int64 in llama4_rope and qwen2vl_mrope - #1343
Open
akshansh47 wants to merge 1 commit into
Open
Cast tl.program_id(0) to int64 in llama4_rope and qwen2vl_mrope#1343akshansh47 wants to merge 1 commit into
akshansh47 wants to merge 1 commit into
Conversation
Same int32 overflow hazard fixed for rope.py / rms_norm.py in linkedin#804. llama4_rope and qwen2vl_mrope missed the cast; Ascend llama4_rope already has it. Fixes linkedin#1335. Co-authored-by: Cursor <cursoragent@cursor.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
#804int64tl.program_id(0)cast to two kernels that missed it:llama4_rope.pyandqwen2vl_mrope.py.rope.py/rms_norm.py; Ascendllama4_ropealready had.to(tl.int64).Test plan
test/transformers/test_llama4_rope.pyandtest_qwen2vl_mrope.pyon GPU CI#804patternFixes #1335
Made with Cursor