[Excutorch][Llama] Decouple input sequence length from kv cache context length #7927
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
…xt length Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/) [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/7927
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 1070600 with merge base bdd3d9c ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
facebook-github-bot
added
the
CLA Signed
This was referenced Jan 24, 2025
kimishpatel
added a commit
that referenced
this pull request
Jan 24, 2025
…xt length Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/) ghstack-source-id: 262854267 Pull Request resolved: #7927
|
This pull request was exported from Phabricator. Differential Revision: D68448334 |
…cache context length" Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/) [ghstack-poisoned]
|
This pull request was exported from Phabricator. Differential Revision: D68448334 |
kimishpatel
added a commit
that referenced
this pull request
Jan 24, 2025
…xt length Pull Request resolved: #7927 Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/) ghstack-source-id: 262945661
…cache context length" Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/) [ghstack-poisoned]
|
This pull request was exported from Phabricator. Differential Revision: D68448334 |
kimishpatel
added a commit
that referenced
this pull request
Jan 24, 2025
…xt length Pull Request resolved: #7927 Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. ghstack-source-id: 263000137 Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/)
…cache context length" Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/) [ghstack-poisoned]
|
This pull request was exported from Phabricator. Differential Revision: D68448334 |
kimishpatel
added a commit
that referenced
this pull request
Jan 27, 2025
…xt length Pull Request resolved: #7927 Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. ghstack-source-id: 263237442 Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/)
…cache context length" Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/) [ghstack-poisoned]
kimishpatel
added a commit
that referenced
this pull request
Jan 28, 2025
…xt length Pull Request resolved: #7927 Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. ghstack-source-id: 263342053 Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/)
|
This pull request was exported from Phabricator. Differential Revision: D68448334 |
…cache context length" Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/) [ghstack-poisoned]
kimishpatel
added a commit
that referenced
this pull request
Jan 28, 2025
…xt length Pull Request resolved: #7927 Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. ghstack-source-id: 263366616 Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/)
|
This pull request was exported from Phabricator. Differential Revision: D68448334 |
kimishpatel
added
module: examples
…cache context length" Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/) cc mergennachin cccclai helunwencser dvorjackz [ghstack-poisoned]
|
This pull request was exported from Phabricator. Differential Revision: D68448334 |
kimishpatel
added a commit
that referenced
this pull request
Jan 28, 2025
…xt length Pull Request resolved: #7927 Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. ghstack-source-id: 263491763 Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/)
…cache context length" Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/) cc mergennachin cccclai helunwencser dvorjackz [ghstack-poisoned]
|
This pull request was exported from Phabricator. Differential Revision: D68448334 |
kimishpatel
added a commit
that referenced
this pull request
Jan 28, 2025
…xt length Pull Request resolved: #7927 Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. ghstack-source-id: 263517976 Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/)
…cache context length" Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/) cc mergennachin cccclai helunwencser dvorjackz [ghstack-poisoned]
|
This pull request was exported from Phabricator. Differential Revision: D68448334 |
kimishpatel
added a commit
that referenced
this pull request
Jan 28, 2025
…xt length Pull Request resolved: #7927 Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. ghstack-source-id: 263531315 Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/)
…cache context length" Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/) cc mergennachin cccclai helunwencser dvorjackz [ghstack-poisoned]
|
This pull request was exported from Phabricator. Differential Revision: D68448334 |
kimishpatel
added a commit
that referenced
this pull request
Jan 29, 2025
…xt length Pull Request resolved: #7927 Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. ghstack-source-id: 263580354 Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/)
…cache context length" Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/) cc mergennachin cccclai helunwencser dvorjackz [ghstack-poisoned]
|
This pull request was exported from Phabricator. Differential Revision: D68448334 |
kimishpatel
added a commit
that referenced
this pull request
Jan 29, 2025
…xt length Pull Request resolved: #7927 Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. ghstack-source-id: 263653316 Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/)
iseeyuan
approved these changes
Jan 30, 2025
facebook-github-bot
merged commit ecb6050
into
gh/kimishpatel/152/base
74 of 77 checks passed
facebook-github-bot
temporarily deployed
to
cherry-pick-bot
GitHub Actions
Inactive
manuelcandales
pushed a commit
that referenced
this pull request
Jan 30, 2025
…xt length (#8047) Pull Request resolved: #7927 Decouple max sequence length, for shape dynamism in torch.export, from sequence length used for kv cache sizing. ghstack-source-id: 263653316 Differential Revision: [D68448334](https://our.internmc.facebook.com/intern/diff/D68448334/) Co-authored-by: Kimish Patel <[email protected]>
kimishpatel
added a commit
that referenced
this pull request
Feb 11, 2025
Summary: Previous PR #7927 deecoupled max_seq_length from kv cache. That broke perf ci workflow. Fix that. Test Plan: Trigger it manually and check Reviewers: Subscribers: Tasks: Tags: [ghstack-poisoned]
kimishpatel
added a commit
that referenced
this pull request
Feb 12, 2025
Summary: Previous PR #7927 deecoupled max_seq_length from kv cache. That broke perf ci workflow. Fix that. Test Plan: Trigger it manually and check apple perf: https://github.com/pytorch/executorch/actions/runs/13267110949 android perf: https://github.com/pytorch/executorch/actions/runs/13267110908 Reviewers: Subscribers: Tasks: Tags: cc guangy10 huydhn kirklandsign shoumikhin [ghstack-poisoned]
kimishpatel
added a commit
that referenced
this pull request
Feb 12, 2025
* [Executorch][perf-ci] Fix perf ci Summary: Previous PR #7927 deecoupled max_seq_length from kv cache. That broke perf ci workflow. Fix that. Test Plan: Trigger it manually and check Reviewers: Subscribers: Tasks: Tags: [ghstack-poisoned] * Update on "[Executorch][perf-ci] Fix perf ci" Summary: Previous PR #7927 deecoupled max_seq_length from kv cache. That broke perf ci workflow. Fix that. Test Plan: Trigger it manually and check apple perf: https://github.com/pytorch/executorch/actions/runs/13267110949 android perf: https://github.com/pytorch/executorch/actions/runs/13267110908 Reviewers: Subscribers: Tasks: Tags: cc guangy10 huydhn kirklandsign shoumikhin [ghstack-poisoned]
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.
Stack from ghstack (oldest at bottom):
Decouple max sequence length, for shape dynamism in torch.export, from sequence
length used for kv cache sizing.
Differential Revision: D68448334
cc @mergennachin @cccclai @helunwencser @dvorjackz