[Executorch][llama] Refactoring sdpa#5667
Conversation
- Change variable name from is_sdpa_with_kv_cache to is_seq_at_dim_1 to be more meaningful - remove seq_len as it can be derived from the size of query Differential Revision: [D62623243](https://our.internmc.facebook.com/intern/diff/D62623243/) [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/5667
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 57b7dd8 with merge base ee32848 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
|
This pull request was exported from Phabricator. Differential Revision: D62623243 |
- Change variable name from is_sdpa_with_kv_cache to is_seq_at_dim_1 to be more meaningful - remove seq_len as it can be derived from the size of query Differential Revision: [D62623243](https://our.internmc.facebook.com/intern/diff/D62623243/) [ghstack-poisoned]
|
This pull request was exported from Phabricator. Differential Revision: D62623243 |
- Change variable name from is_sdpa_with_kv_cache to is_seq_at_dim_1 to be more meaningful - remove seq_len as it can be derived from the size of query Differential Revision: [D62623243](https://our.internmc.facebook.com/intern/diff/D62623243/) [ghstack-poisoned]
|
This pull request was exported from Phabricator. Differential Revision: D62623243 |
- Change variable name from is_sdpa_with_kv_cache to is_seq_at_dim_1 to be more meaningful - remove seq_len as it can be derived from the size of query Differential Revision: [D62623243](https://our.internmc.facebook.com/intern/diff/D62623243/) [ghstack-poisoned]
|
This pull request was exported from Phabricator. Differential Revision: D62623243 |
- Change variable name from is_sdpa_with_kv_cache to is_seq_at_dim_1 to be more meaningful - remove seq_len as it can be derived from the size of query Differential Revision: [D62623243](https://our.internmc.facebook.com/intern/diff/D62623243/) [ghstack-poisoned]
|
This pull request was exported from Phabricator. Differential Revision: D62623243 |
- Change variable name from is_sdpa_with_kv_cache to is_seq_at_dim_1 to be more meaningful - remove seq_len as it can be derived from the size of query Differential Revision: [D62623243](https://our.internmc.facebook.com/intern/diff/D62623243/) [ghstack-poisoned]
|
This pull request was exported from Phabricator. Differential Revision: D62623243 |
- Change variable name from is_sdpa_with_kv_cache to is_seq_at_dim_1 to be more meaningful - remove seq_len as it can be derived from the size of query Differential Revision: [D62623243](https://our.internmc.facebook.com/intern/diff/D62623243/) [ghstack-poisoned]
|
This pull request was exported from Phabricator. Differential Revision: D62623243 |
|
This pull request has been merged in 29364c4. |
Pull Request resolved: pytorch/executorch#5667 - Change variable name from is_sdpa_with_kv_cache to is_seq_at_dim_1 to be more meaningful - remove seq_len as it can be derived from the size of query ghstack-source-id: 245751540 @exported-using-ghexport Differential Revision: [D62623243](https://our.internmc.facebook.com/intern/diff/D62623243/)
Stack from ghstack (oldest at bottom):
Differential Revision: D62623243