[TE/JAX] XLA FFI calls for Softmax and FusedAttnBackward #1319
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
|
/te-ci jax L1 |
|
@zlsh80826 The forward test in |
Signed-off-by: Hua Huang <huah@nvidia.com>
FusedAttnBackward passed all testes in test_fused_attn.py. Dequantize is not used currently; finish it for completeness. Signed-off-by: Hua Huang <huah@nvidia.com>
Signed-off-by: Hua Huang <huah@nvidia.com>
for more information, see https://pre-commit.ci
|
CI L1 tests passed. Rebase to the main branch to verify #1314 |
huanghua1994
force-pushed
the
xla-ffi-softmax-attnbwd
branch
from
73e00ba
to
e891678
Compare
|
/te-ci jax L1 |
zlsh80826
reviewed
Nov 9, 2024
Signed-off-by: Hua Huang <huah@nvidia.com>
|
/te-ci jax L1 |
zlsh80826
approved these changes
Nov 11, 2024
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Outdated suggestions cannot be applied.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.
Description
This PR introduced the following primitives implemented with the new custom calls:
ScaledSoftmaxFwdPrimitiveScaledSoftmaxBwdPrimitiveScaledMaskedSoftmaxFwdPrimitiveScaledMaskedSoftmaxBwdPrimitiveScaledUpperTriangMaskedSoftmaxFwdPrimitiveScaledUpperTriangMaskedSoftmaxBwdPrimitiveFusedAttnBwdPrimitiveAlso added
DequantizeFFI()intransformer_engine/jax/csrc/extensions/quantization.cppalthough currently no Python function calls dequantize explicitly.All C++ functions in
transformer_engine/jax/csrc/extensionshave FFI after this PR.Type of change
Checklist: