Commits · ca81f32e04b3dcab69ff919322abab578a26d12f · gaoqiong / flash-attention

05 Nov, 2022 1 commit
- Implement rotary embedding in CUDA · ca81f32e
  Tri Dao authored Nov 04, 2022
  
  ca81f32e
04 Nov, 2022 3 commits
- Fix more race condition in Triton bwd when there's bias · 62025e1a
  Tri Dao authored Nov 04, 2022
  
  62025e1a
- Fix race condition in Triton bwd when there's bias · ff78ea41
  Tri Dao authored Nov 04, 2022
  
  ff78ea41
- Implement attention bias for Triton version · 86862cfd
  Tri Dao authored Nov 04, 2022
  
  86862cfd
03 Nov, 2022 1 commit
- Fix race condition for Triton bwd for headdim 48 and 96 · 470010f5
  Tri Dao authored Nov 03, 2022
  
  470010f5
02 Nov, 2022 1 commit
- Fix race condition in Triton bwd for non-po2 headdims · aacc10fb
  Tri Dao authored Nov 02, 2022
  
  aacc10fb
01 Nov, 2022 2 commits
- Avoid memcpy in the Triton bwd · 1fb12afd
  Tri Dao authored Nov 01, 2022
  
  1fb12afd
- Fix race conditions in the Triton bwd for headdim=64 · 731f154d
  Tri Dao authored Nov 01, 2022
  
  731f154d
31 Oct, 2022 10 commits
- Fix race condition in Triton fwd · 9b0bc978
  Tri Dao authored Oct 31, 2022
  
  9b0bc978
- Fix EVEN_M & EVEN_HEADDIM for headdim=40 in Triton bwd · 215930bc
  Tri Dao authored Oct 31, 2022
  
  215930bc
- Add debug_barrier for all headdims in Triton bwd · 4f81aff4
  Tri Dao authored Oct 31, 2022
  
  4f81aff4
- Disable some autotune configs that give wrong results in Triton bwd · bedcbd6a
  Tri Dao authored Oct 31, 2022
  
  bedcbd6a
- [WIP] Support all head dimensions up to 128 in the Triton bwd · e78d509c
  Tri Dao authored Oct 31, 2022
```
WIP because there seems to be some race conditions for head dimensions other
than 16, 32, 64, 128.
```
  e78d509c
- Support all head dimensions up to 128 in the Triton fwd · 008951f1
  Tri Dao authored Oct 30, 2022
  
  008951f1
- Support arbitrary seqlens (both q & k) in Triton bwd · b910bf14
  Tri Dao authored Oct 30, 2022
  
  b910bf14
- Support arbitrary seqlen_k in Triton bwd · dc554693
  Tri Dao authored Oct 30, 2022
  
  dc554693
- Fix Triton fwd to support seqlen not multiples of 128 · d11341fd
  Tri Dao authored Oct 30, 2022
  
  d11341fd
- Implement FlashAttention in Triton · b0c0db81
  Tri Dao authored Oct 30, 2022
  
  b0c0db81
25 Oct, 2022 1 commit
- Get rid of o_rows_are_valid since we don't have headdim=16 anymore · c422fee3
  Tri Dao authored Oct 24, 2022
  
  c422fee3
24 Oct, 2022 4 commits
- Support all head dims that are multiples of 8, up to 128 · 46fd2a20
  Tri Dao authored Oct 24, 2022
  
  46fd2a20
- Cast q.get_device() to char to avoid compiler warning (narrowing) · 97e13de2
  Tri Dao authored Oct 24, 2022
  
  97e13de2
- Add Megatron attention implementation for benchmarking · ed553e92
  Tri Dao authored Oct 23, 2022
  
  ed553e92
- Add Triton implementation for benchmarking · 50ca2348
  Tri Dao authored Oct 23, 2022
  
  50ca2348
23 Oct, 2022 4 commits
- Attempt to use atomicCAS to replace atomicAdd(bfloat16) · 9e92a1f2
  Tri Dao authored Oct 23, 2022
  
  9e92a1f2
- Merge pull request #61 from robotcator/workflow · 6731855b
  Tri Dao authored Oct 23, 2022
```
build wheel and upload to release
```
  6731855b
- Move benchmark utils, support AMP · fb88e5e4
  Tri Dao authored Oct 23, 2022
  
  fb88e5e4
- Split bwd on the seqlen_q dimension · a5a8806d
  Tri Dao authored Oct 23, 2022
  
  a5a8806d
22 Oct, 2022 1 commit
- Don't need to run configure for the forward pass · 871db479
  Tri Dao authored Oct 21, 2022
  
  871db479
21 Oct, 2022 3 commits
- Use block_size=128 for headdim=128 on SM80 · 7fc39832
  Tri Dao authored Oct 21, 2022
```
Previously we were using block_size=256.
```
  7fc39832
- Split fwd on the seqlen_q dimension · a44f48df
  Tri Dao authored Oct 21, 2022
  
  a44f48df
- Rework dropout to decouple forward and backward · 1aa6d7d9
  Tri Dao authored Oct 18, 2022
```
They don't have to have the same block size, number of threads, etc.
```
  1aa6d7d9
17 Oct, 2022 4 commits
- Merge pull request #60 from 201419/patch-1 · 1d0b41be
  Tri Dao authored Oct 17, 2022
```
fix typo in function mha_fwd
```
  1d0b41be
- Merge branch 'main' of github.com:robotcator/flash-attention into workflow · 35d589fa
  robotcator authored Oct 17, 2022
  
  35d589fa
- using tag trigger rather than push trigger · 10d07459
  robotcator authored Oct 17, 2022
  
  10d07459
- fix typo in function mha_fwd · ff07250e
  YangShu authored Oct 17, 2022
```
as title.
```
  ff07250e
16 Oct, 2022 1 commit
- Fix #54: set device for multi-GPU case · 52fb4b72
  Tri Dao authored Oct 16, 2022
  
  52fb4b72
14 Oct, 2022 2 commits
- Fix QKV interface to allocate output in Python · 1b9facac
  Tri Dao authored Oct 14, 2022
  
  1b9facac
- Implement attention kernel that splits the batch into two · 5badfb78
  Tri Dao authored Oct 13, 2022
  
  5badfb78
10 Oct, 2022 1 commit
- Merge pull request #53 from robotcator/workflow · f515c77f
  Tri Dao authored Oct 09, 2022
```
build wheel workflow
```
  f515c77f
06 Oct, 2022 1 commit
- Merge pull request #55 from ajfadam/main · 8dd52b07
  Tri Dao authored Oct 06, 2022
```
remove numpy dependency
```
  8dd52b07