Fix cache_positions tensor size in TextLLMRunner #12476
Merged
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
In Huggingface causal LM forward convention, `cache_position` should have the same length as `input_ids`. The previous logic will allocate `cache_position` based on method metadata which by default is equal to the maximum length of this tensor (normally max context length). Now changing the logic to align the size of `cache_position` to `input_ids`. [ghstack-poisoned]
larryliu0820
requested review from
jackzhxng,
mergennachin and
swolchok
as code owners
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/12476
Note: Links to docs will display an error until the docs builds have been completed. ⏳ No Failures, 6 PendingAs of commit efd7096 with merge base 042eb1a ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
facebook-github-bot
added
the
CLA Signed
This PR needs a
|
lucylq
approved these changes
Jul 15, 2025
larryliu0820
temporarily deployed
to
cherry-pick-bot
GitHub Actions
Inactive
larryliu0820
added a commit
that referenced
this pull request
Jul 15, 2025
This PR was created by the merge bot to help merge the original PR into the main branch. ghstack PR number: #12476 by @larryliu0820 ^ Please use this as the source of truth for the PR details, comments, and reviews ghstack PR base: https://github.com/pytorch/executorch/tree/gh/larryliu0820/68/base ghstack PR head: https://github.com/pytorch/executorch/tree/gh/larryliu0820/68/head Merge bot PR base: https://github.com/pytorch/executorch/tree/main Merge bot PR head: https://github.com/pytorch/executorch/tree/gh/larryliu0820/68/orig @diff-train-skip-merge Co-authored-by: Mengwei Liu <[email protected]>
pytorchbot
added a commit
that referenced
this pull request
Jul 15, 2025
This PR was created by the merge bot to help merge the original PR into the main branch. ghstack PR number: #12476 by @larryliu0820 ^ Please use this as the source of truth for the PR details, comments, and reviews ghstack PR base: https://github.com/pytorch/executorch/tree/gh/larryliu0820/68/base ghstack PR head: https://github.com/pytorch/executorch/tree/gh/larryliu0820/68/head Merge bot PR base: https://github.com/pytorch/executorch/tree/main Merge bot PR head: https://github.com/pytorch/executorch/tree/gh/larryliu0820/68/orig @diff-train-skip-merge Co-authored-by: Mengwei Liu <[email protected]> (cherry picked from commit aa44c06)
psiddh
added a commit
that referenced
this pull request
Jul 16, 2025
This PR was created by the merge bot to help merge the original PR into the main branch. ghstack PR number: #12476 by @larryliu0820 ^ Please use this as the source of truth for the PR details, comments, and reviews ghstack PR base: https://github.com/pytorch/executorch/tree/gh/larryliu0820/68/base ghstack PR head: https://github.com/pytorch/executorch/tree/gh/larryliu0820/68/head Merge bot PR base: https://github.com/pytorch/executorch/tree/main Merge bot PR head: https://github.com/pytorch/executorch/tree/gh/larryliu0820/68/orig @diff-train-skip-merge Co-authored-by: Mengwei Liu <[email protected]> Co-authored-by: Siddartha Pothapragada <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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.
Stack from ghstack (oldest at bottom):
In Huggingface causal LM forward convention,
cache_positionshouldhave the same length as
input_ids. The previous logic will allocatecache_positionbased on method metadata which by default is equal tothe maximum length of this tensor (normally max context length). Now
changing the logic to align the size of
cache_positiontoinput_ids.