[dtensor] run transformer sdpa in dtensor #122997
Closed
Conversation
This file contains 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
[ghstack-poisoned]
This was referenced Mar 29, 2024
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/122997
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit b336a91 with merge base a3d97f6 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
pytorch-bot
bot
added
the
oncall: distributed
tianyu-l
added a commit
that referenced
this pull request
Mar 29, 2024
ghstack-source-id: 661b23ba34e8a05ca8eb3057fe04b82832ad64d1 Pull Request resolved: #122997
tianyu-l
added
ciflow/trunk
Now that efficient attention is supported in dtensor, we can modify the transformer test to use dtensor in SDPA and get rid of the manual num_head adjustments. Caveat: Efficient attention is supported only with bf16/fp32 (not fp64) and has other constraints. If any of the constraints are not satisfied, the SDPA would fall back to the math decomposed attention, which will break as it does not fully work with dtensor (it creates a `torch.Tensor` mask in the middle). I considered adding some checks like in P1202254918 but that needs to be added everywhere this Transformer is used. Is it necessary if the current CI machines can run efficient attention? Test files containing this Transformer: - `test/distributed/tensor/parallel/test_tp_examples.py` - `test/distributed/_composable/fsdp/test_fully_shard_training.py` - `test/distributed/_composable/fsdp/test_fully_shard_clip_grad_norm_.py` cc mrshenli pritamdamania87 zhaojuanmao satgera rohan-varma gqchen aazzolini osalpekar jiayisuse H-Huang kwen2501 awgu penguinwu fegin XilunWu wanchaol fduwjj wz337 wconstab yf225 chauhang [ghstack-poisoned]
tianyu-l
added a commit
that referenced
this pull request
May 6, 2024
ghstack-source-id: f81c79a4dd2d78e33b28f445c9a10bebb1a80997 Pull Request resolved: #122997
Now that efficient attention is supported in dtensor, we can modify the transformer test to use dtensor in SDPA and get rid of the manual num_head adjustments. Caveat: Efficient attention is supported only with bf16/fp32 (not fp64) and has other constraints. If any of the constraints are not satisfied, the SDPA would fall back to the math decomposed attention, which will break as it does not fully work with dtensor (it creates a `torch.Tensor` mask in the middle). I considered adding some checks like in P1202254918 but that needs to be added everywhere this Transformer is used. Is it necessary if the current CI machines can run efficient attention? Test files containing this Transformer: - `test/distributed/tensor/parallel/test_tp_examples.py` - `test/distributed/_composable/fsdp/test_fully_shard_training.py` - `test/distributed/_composable/fsdp/test_fully_shard_clip_grad_norm_.py` cc mrshenli pritamdamania87 zhaojuanmao satgera rohan-varma gqchen aazzolini osalpekar jiayisuse H-Huang kwen2501 awgu penguinwu fegin XilunWu wanchaol fduwjj wz337 wconstab yf225 chauhang [ghstack-poisoned]
tianyu-l
added a commit
that referenced
this pull request
May 6, 2024
ghstack-source-id: 1bc4d914853e4d647c72f37485b5b1102771a23e Pull Request resolved: #122997
Now that efficient attention is supported in dtensor, we can modify the transformer test to use dtensor in SDPA and get rid of the manual num_head adjustments. Caveat: Efficient attention is supported only with bf16/fp32 (not fp64) and has other constraints. If any of the constraints are not satisfied, the SDPA would fall back to the math decomposed attention, which will break as it does not fully work with dtensor (it creates a `torch.Tensor` mask in the middle). I considered adding some checks like in P1202254918 but that needs to be added everywhere this Transformer is used. Is it necessary if the current CI machines can run efficient attention? Test files containing this Transformer: - `test/distributed/tensor/parallel/test_tp_examples.py` - `test/distributed/_composable/fsdp/test_fully_shard_training.py` - `test/distributed/_composable/fsdp/test_fully_shard_clip_grad_norm_.py` cc mrshenli pritamdamania87 zhaojuanmao satgera rohan-varma gqchen aazzolini osalpekar jiayisuse H-Huang kwen2501 awgu penguinwu fegin XilunWu wanchaol fduwjj wz337 wconstab yf225 chauhang [ghstack-poisoned]
tianyu-l
added a commit
that referenced
this pull request
May 7, 2024
ghstack-source-id: bf6e6601fc4b03adf9fbe34cbb3e5ad574746665 Pull Request resolved: #122997
Now that efficient attention is supported in dtensor, we can modify the transformer test to use dtensor in SDPA and get rid of the manual num_head adjustments. Caveat: Efficient attention is supported only with bf16/fp32 (not fp64) and has other constraints. If any of the constraints are not satisfied, the SDPA would fall back to the math decomposed attention, which will break as it does not fully work with dtensor (it creates a `torch.Tensor` mask in the middle). I considered adding some checks like in P1202254918 but that needs to be added everywhere this Transformer is used. Is it necessary if the current CI machines can run efficient attention? Test files containing this Transformer: - `test/distributed/tensor/parallel/test_tp_examples.py` - `test/distributed/_composable/fsdp/test_fully_shard_training.py` - `test/distributed/_composable/fsdp/test_fully_shard_clip_grad_norm_.py` cc mrshenli pritamdamania87 zhaojuanmao satgera rohan-varma gqchen aazzolini osalpekar jiayisuse H-Huang kwen2501 awgu penguinwu fegin XilunWu wanchaol fduwjj wz337 wconstab yf225 chauhang [ghstack-poisoned]
tianyu-l
added a commit
that referenced
this pull request
May 7, 2024
ghstack-source-id: 2edbf1c3b1288f5ec404eafa3087e4964b222578 Pull Request resolved: #122997
Now that efficient attention is supported in dtensor, we can modify the transformer test to use dtensor in SDPA and get rid of the manual num_head adjustments. Caveat: Efficient attention is supported only with bf16/fp32 (not fp64) and has other constraints. If any of the constraints are not satisfied, the SDPA would fall back to the math decomposed attention, which will break as it does not fully work with dtensor (it creates a `torch.Tensor` mask in the middle). I considered adding some checks like in P1202254918 but that needs to be added everywhere this Transformer is used. Is it necessary if the current CI machines can run efficient attention? Test files containing this Transformer: - `test/distributed/tensor/parallel/test_tp_examples.py` - `test/distributed/_composable/fsdp/test_fully_shard_training.py` - `test/distributed/_composable/fsdp/test_fully_shard_clip_grad_norm_.py` cc mrshenli pritamdamania87 zhaojuanmao satgera rohan-varma gqchen aazzolini osalpekar jiayisuse H-Huang kwen2501 awgu penguinwu fegin XilunWu wanchaol fduwjj wz337 wconstab yf225 chauhang [ghstack-poisoned]
tianyu-l
added a commit
that referenced
this pull request
May 7, 2024
ghstack-source-id: 514e047b05af916c1a23ec4e6b45338a976eae07 Pull Request resolved: #122997
Now that efficient attention is supported in dtensor, we can modify the transformer test to use dtensor in SDPA and get rid of the manual num_head adjustments. Caveat: Efficient attention is supported only with bf16/fp32 (not fp64) and has other constraints. If any of the constraints are not satisfied, the SDPA would fall back to the math decomposed attention, which will break as it does not fully work with dtensor (it creates a `torch.Tensor` mask in the middle). I considered adding some checks like in P1202254918 but that needs to be added everywhere this Transformer is used. Is it necessary if the current CI machines can run efficient attention? Test files containing this Transformer: - `test/distributed/tensor/parallel/test_tp_examples.py` - `test/distributed/_composable/fsdp/test_fully_shard_training.py` - `test/distributed/_composable/fsdp/test_fully_shard_clip_grad_norm_.py` cc mrshenli pritamdamania87 zhaojuanmao satgera rohan-varma gqchen aazzolini osalpekar jiayisuse H-Huang kwen2501 awgu penguinwu fegin XilunWu wanchaol fduwjj wz337 wconstab yf225 chauhang [ghstack-poisoned]
tianyu-l
added a commit
that referenced
this pull request
May 8, 2024
ghstack-source-id: 9a2cac1eaba0211bd115a00db2aa5eded8fae36a Pull Request resolved: #122997
|
@pytorchbot merge |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
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):
Now that efficient attention is supported in dtensor, we can modify the transformer test to use dtensor in SDPA and get rid of the manual num_head adjustments.
Caveat: Efficient attention is supported only with bf16/fp32 (not fp64) and has other constraints. If any of the constraints are not satisfied, the SDPA would fall back to the math decomposed attention, which will break as it does not fully work with dtensor (it creates a
torch.Tensormask in the middle). I considered adding some checks like in P1202254918 but that needs to be added everywhere this Transformer is used. Is it necessary if the current CI machines can run efficient attention?Test files containing this Transformer:
test/distributed/tensor/parallel/test_tp_examples.pytest/distributed/_composable/fsdp/test_fully_shard_training.pytest/distributed/_composable/fsdp/test_fully_shard_clip_grad_norm_.pycc @mrshenli @pritamdamania87 @zhaojuanmao @satgera @gqchen @aazzolini @osalpekar @jiayisuse @H-Huang @kwen2501 @awgu @penguinwu @fegin @XilunWu @wanchaol @fduwjj @wz337 @wconstab @yf225 @chauhang @d4l3k @rohan-varma