Commits · fdc09f42cc1532658a333eb34fff7e683dc2b2e8 · OpenDAS / TransformerEngine

27 Jul, 2023 1 commit

Exposing RMSNorm in pyTorch (#306) · fdc09f42

Przemyslaw Tredak authored Jul 27, 2023



* Exposing RMSNorm in pyTorch extensions
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

* First pass at the Python API
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

* Small fixes
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

* Added numerics tests and fixed issues
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

* Lint fixes
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

* Added RMSNorm to LayerNormMLP
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

* Added ONNX export and tests for RMSNorm
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

* Fix python lint
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

* Fix BERT case
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

* Added normalization option to the TransformerLayer
Added tests
Fixed test failures
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

* Fix documentation
Co-authored-by: Przemyslaw Tredak <ptrendx@gmail.com>
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Fix kwarg bug
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Fix IMA and invalid type error
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Increase RMSNorm threshold for bf16 case
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Fix ONNX tests
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

---------
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

fdc09f42

26 Jul, 2023 1 commit
- Make QK layer scaling opt-in (#339) · 06d5fa97
  Kirthi Shankar Sivamani authored Jul 25, 2023
```
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
```
  06d5fa97
20 Jul, 2023 4 commits

[GH Actions] Build PyTorch without FA for memory usage (#334) · 7a30ba45
Kirthi Shankar Sivamani authored Jul 20, 2023
```
Build PyTorch with FA for memory usage
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
```
7a30ba45

[Paddle] Support flash attention (#330) · 749a7f69

Shijie authored Jul 20, 2023



* add flash attn tests
Signed-off-by: Shijie Wang <jaywan@nvidia.com>

* update flash attn
Signed-off-by: Shijie Wang <jaywan@nvidia.com>

* fix random seed
Signed-off-by: Shijie Wang <jaywan@nvidia.com>

---------
Signed-off-by: Shijie Wang <jaywan@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

749a7f69

Replace deprecated sharding API in JAX test (#332) · 23f4864d

Tim Moon authored Jul 19, 2023



Replace deprecated sharding API
Signed-off-by: Tim Moon <tmoon@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

23f4864d

Relax FA 2.0 checks for Ada (#331) · f0ddab82
Kirthi Shankar Sivamani authored Jul 19, 2023
```
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
```
f0ddab82

19 Jul, 2023 4 commits

[Paddle] Add nn Layers (BF16) (#299) · 10eb13e2

Tian Zheng authored Jul 20, 2023



* Add Linear layer (FP16)
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>

- Add BF16 training example
- Add fp8_autocast (only supports non-fp8 for now)
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>

* Remove FP8 stuff
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>

* Simplify Linear layer forward
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>

* Add LayerNorm layer (BF16)
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>

* Add LayerNormLinear layer
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>

* Store weights in BF16
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>

* Add LayerNormMLP layer
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>

* Add BF16 MNIST example
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>

* Remove in-place cast for compatibility with Paddle AMP mechanism
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>

* README correction
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>

* Add Paddle op as a backend option
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>

* Fix code format
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>

* Fix dtype change between iterations
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>

* Minor fixes
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>

* Move forward function out of base layer
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>

* Use Paddle nvtx bindings
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>

---------
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>

10eb13e2

FlashAttention 2.0 support (#329) · faefacb8

Kirthi Shankar Sivamani authored Jul 18, 2023



* FA v2.0 support
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* fix typo
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

---------
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

faefacb8

Fix numerics for activation recompute (#327) · ded8b9bd
Kirthi Shankar Sivamani authored Jul 18, 2023
```
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
```
ded8b9bd

FA does not support head_dim > 64 on Ada (#328) · 2a81e939

Kirthi Shankar Sivamani authored Jul 18, 2023



* FA does not support head_dim > 64 on Ada
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Add cc8.7 to no FA list
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

---------
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

2a81e939

18 Jul, 2023 2 commits

[JAX] Fully remove attn_type and set self_attn_mask_type default to 'causal' (#324) · a3e4e611

zlsh80826 authored Jul 18, 2023



* Fully remove attn_type and set self_attn_mask_type default to 'causal'
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Fix tests with new arguments
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Explicit self_attn_mask_type for examples
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Update transformer_engine/jax/flax/transformer.py
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Signed-off-by: zlsh80826 <rewang@nvidia.com>

* Update transformer_engine/jax/flax/transformer.py
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Signed-off-by: zlsh80826 <rewang@nvidia.com>

---------
Signed-off-by: Reese Wang <rewang@nvidia.com>
Signed-off-by: zlsh80826 <rewang@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

a3e4e611

layer number bug fixes (#326) · 21162d06

Kirthi Shankar Sivamani authored Jul 18, 2023



* layer number bug fixes
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Fix
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

---------
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

21162d06

17 Jul, 2023 1 commit
- Change version to 0.12.0dev · 706d7da3
  Kirthi Shankar Sivamani authored Jul 17, 2023
```
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
```
  706d7da3
15 Jul, 2023 2 commits

Disable TorchDynamo optimizations in PyTorch modules (#312) · c6538d6e

Tim Moon authored Jul 15, 2023



* Disable TorchDynamo optimizations in PyTorch modules
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Add test for Torch Dynamo
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Add torch.dynamo test to qa
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Skip torch.compile test for <v2.0
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

---------
Signed-off-by: Tim Moon <tmoon@nvidia.com>
Signed-off-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

c6538d6e

Stop creating pip wheels in GitHub build tests (#315) · 36873ec8

Tim Moon authored Jul 15, 2023



* Stop creating pip wheels in GitHub build tests
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Revert to 23.03 container in GitHub build tests
Signed-off-by: Tim Moon <tmoon@nvidia.com>

---------
Signed-off-by: Tim Moon <tmoon@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

36873ec8

14 Jul, 2023 4 commits

[PyTorch] Fix FP8 checkpointing for non forward execution cases (#323) · f351191b
Kirthi Shankar Sivamani authored Jul 14, 2023
```
Bug fix for checkpointing
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
```
f351191b

Fix calibration in PyTorch example (#322) · 04822f40

Kirthi Shankar Sivamani authored Jul 14, 2023



* Fix example
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Review
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

---------
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

04822f40

Fused attention fixes for cuDNN 8.9.3 (#311) · 0707552e

cyanguwa authored Jul 14, 2023



* Fix bprop for cuDNN 8.9.3
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* Update cuDNN version requirement to 8.9.3
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* debug paddle CI
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* debug paddle CI; force LD_LIBRARY
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* debug paddle CI; force LD_LIBRARY to /opt
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* remove debug info for paddle
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* change cudnn requirement to 8.9.1 for v1 and 8.9.0 for v2; add batch size 32 for unit test; add LD library path for paddle tests temporarily
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* remove printf line in fused_attn.cpp
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* add batch size 32 for unit test
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* update cudnn-frontend to 0.9.2
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* remove temporary LD library path used for testing pre-released cudnn 8.9.3
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

---------
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

0707552e

Deprecate unused APIs (#321) · 58d2ebab

Kirthi Shankar Sivamani authored Jul 14, 2023



* Deprecate unused APIs
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* review comments
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Review
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

---------
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

58d2ebab

13 Jul, 2023 4 commits

Fix dtype for KV inference cache (#319) · b172bad8
Kirthi Shankar Sivamani authored Jul 13, 2023
```
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
```
b172bad8

Fix FP32 LayerNorm ONNX export (#313) · 6bd35bf9

Neta Zmora authored Jul 13, 2023



* Fix FP32 LayerNorm ONNX export

When running inference use a fwd method that is registered with torchscript.
Signed-off-by: Neta Zmora <nzmora@nvidia.com>

* Bug fix
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

---------
Signed-off-by: Neta Zmora <nzmora@nvidia.com>
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

6bd35bf9

Remove buffer registration for FSDP like cases (#318) · 11c5d588
Kirthi Shankar Sivamani authored Jul 13, 2023
```
Remove extra buffers
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
```
11c5d588

Catch cublas FP8 errors (#317) · 8c3110d1

Kirthi Shankar Sivamani authored Jul 12, 2023



* Better dimension assert for FP8
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* line
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

---------
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

8c3110d1

07 Jul, 2023 1 commit
- [JAX] Support arbitrary dimensinos of fp8 meta. (#309) · a7bc7cf7
  Ming-Xu Huang authored Jul 07, 2023
```
Signed-off-by: Ming Huang <mingh@nvidia.com>
```
  a7bc7cf7
06 Jul, 2023 1 commit

Add operators for Paddle (#285) · a7a1a070

Shijie authored Jul 07, 2023



* add more ops
Signed-off-by: Shijie Wang <jaywan@nvidia.com>

* add skipif
Signed-off-by: Shijie Wang <jaywan@nvidia.com>

* fix bug
Signed-off-by: Shijie Wang <jaywan@nvidia.com>

* minor change
Signed-off-by: Shijie Wang <jaywan@nvidia.com>

* minor change on coding style
Signed-off-by: Shijie Wang <jaywan@nvidia.com>

---------
Signed-off-by: Shijie Wang <jaywan@nvidia.com>

a7a1a070

01 Jul, 2023 1 commit

Check for cuDNN frontend API when building (#307) · a83605df

Tim Moon authored Jul 01, 2023


Signed-off-by: Tim Moon <tmoon@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

a83605df

30 Jun, 2023 1 commit
- Fix TE discovery in python virtual environments (#303) · 5976fe19
  Tejaswin Parthasarathy authored Jun 30, 2023
```
fix : TE virtuelenv discovery
Signed-off-by: tejaswinp <tejaswinp@nvidia.com>
```
  5976fe19
27 Jun, 2023 1 commit
- Add FP8 paper (#301) · 6f7853dc
  Kirthi Shankar Sivamani authored Jun 27, 2023
```
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
```
  6f7853dc
26 Jun, 2023 3 commits

Get default dtype from pytorch (#300) · 79a78cae

Kirthi Shankar Sivamani authored Jun 26, 2023



* Get default dtype from pytorch
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Review comments
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

---------
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

79a78cae

Test ONNX export - missing BF16 GEMM tests + output.json fix (#297) · 6bccc76e

galagam authored Jun 26, 2023


Signed-off-by: Gal Hubara Agam <ghubaraagam@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

6bccc76e

Fix compute type for GEMM (#296) · c56646e4
Kirthi Shankar Sivamani authored Jun 26, 2023
```
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
```
c56646e4

23 Jun, 2023 1 commit

Fix layer_norm ONNX export (#293) · 574f1b41

Neta Zmora authored Jun 23, 2023



* Fix ONNX export of layer_norm

ONNX has a spec bug: ConstantOfShape supports all dtypes except for BF16.
To WAR we use dtype FP32 and then cast to BF16.

Will also issue a PR to the ONNX sig committee to change the spec in opset 20.
Signed-off-by: Neta Zmora <nzmora@nvidia.com>

* fix lint
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

---------
Signed-off-by: Neta Zmora <nzmora@nvidia.com>
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

574f1b41

22 Jun, 2023 3 commits

Add more TE operators for Paddle (#262) · 4daea528

Tian Zheng authored Jun 23, 2023



* Add cast_transpose

Add gelu, gelu_fp8

Add cast_transpose_bgrad_dgelu

Add layernorm_fwd and layernorm_fwd_fp8

Add layernorm_bwd
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>

* Fix missing header
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>

---------
Signed-off-by: Tian Zheng (Engrg-Hardware 1) <tizheng@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

4daea528

Add long sequence support for fused attention (#237) · 5c58beaa

cyanguwa authored Jun 22, 2023



* add long sequence support and unify three backends for fused attention
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* update cudnn-frontend to v0.9.1
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* replace cpu_float2half_rn with __float2half_rn
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* fix backend selection and NVTEDType
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* minor fixes
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* fix ci
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* make cudnn plan caches thread_local
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* fix CI
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* replace cuDNN throw with NVTE_CHECK
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* fix replacement of cuDNN throw with NVTE_CHECK
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* force dropout probablity to 0 in inference mode
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* change negInfinity to be consistent with m512 fused attn
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* remove float2half conversion for scale_dropout
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* add back runtime api for sm detection
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* add gemm3 to enums FP8Fwd/BwdTensors
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* change dropout from no to yes for fmha_v1
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* remove output_rng_state in m512 kernels
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* fix elts_per_thread calculation in kvpacked fwd
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* remove dropout=0.0 restriction for m512 fused attn
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* remove output_rng_state completely from m512 kernels
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

---------
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

5c58beaa

Update README.rst - Integrations section (#292) · 4330e025

Santosh Bhavani authored Jun 22, 2023



Update README.rst

Added NVIDIA NeMo to Integrations section
Signed-off-by: Santosh Bhavani <santosh.bhavani@live.com>

4330e025

21 Jun, 2023 1 commit
- Fix BF16 ONNX export for successful ONNX Runtime Verification (#290) · 804f1203
  asfiyab-nvidia authored Jun 20, 2023
```
Signed-off-by: Asfiya Baig <asfiyab@nvidia.com>
```
  804f1203
20 Jun, 2023 4 commits

Consistent docs for fuse_wgrad_accumulation (#289) · 0426feb6
Kirthi Shankar Sivamani authored Jun 20, 2023
```
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
```
0426feb6

ONNX export test - fix file name in transformer layer tests (#288) · 918a9ad7

galagam authored Jun 20, 2023

Same filename is used for all activations, file is getting overridden to last activation type
Signed-off-by: Gal Hubara Agam <ghubaraagam@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

918a9ad7

Support dropout for the fused attention when max seqlen <= 512 (#227) · 0816583a

zlsh80826 authored Jun 21, 2023



* Enable fused attention dropout
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Cast the uint32 key/counter to int64
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Update dropout support in fused attention docs
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Revise devPtrCuSeqlen* to align the naming
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Support different Jax PRNG impls
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Revert CastAsync since it is not used
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Implement is_training for 16-bit fused attn
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Add fused attn with dropout sanity unit tests
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Enhance the comments readability and rng_state checker
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Change the attention dropout shape to align other frameworks
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Make encoder tests deterministic
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Change the default seed for the jax encoder tests
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Maintain offset in TE
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Enhance the resource safety
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Revert rng_state type to allow only i64
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Handle the corner case for elts_per_threads calculation
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Populate rng state by kernels
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Rename rng_state as seed in cpp_extensions
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Update the attention dropout comment
Signed-off-by: Reese Wang <rewang@nvidia.com>

---------
Signed-off-by: Reese Wang <rewang@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

0816583a

[JAX] Add self_attn_mask_type and replace attn_type (#273) · 16208b3b

zlsh80826 authored Jun 20, 2023



* Add self_attn_mask_type and replace attn_type
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Refine the keyword style for the better readability
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Replace attn_type with attn_mask_type in praxis transformer
Signed-off-by: Reese Wang <rewang@nvidia.com>

* Fix typos
Signed-off-by: Reese Wang <rewang@nvidia.com>

---------
Signed-off-by: Reese Wang <rewang@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

16208b3b