Commits · 82f60befa3d77253f97d6b054e58a655809db310 · OpenDAS / vllm_cscc

28 May, 2025 1 commit
- set VLLM_FLASH_ATTN_BACKEND to use FlashAttention Backend for attention computation on rocm · 82f60bef
  zhuwenwen authored May 28, 2025
  
  82f60bef
26 May, 2025 1 commit
- fix kernels tests of attention and core · f1467ce5
  zhuwenwen authored May 26, 2025
  
  f1467ce5
07 May, 2025 1 commit
- update v1 mla · ef8f16f4
  zhuwenwen authored May 07, 2025
  
  ef8f16f4
02 May, 2025 1 commit
- [BugFix][Attention] Fix sliding window attention in V1 giving incorrect results (#17574) · 3015d563
  Lucas Wilkinson authored May 02, 2025
```
Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
```
  3015d563
28 Apr, 2025 2 commits

[BugFix] Fix cascade attention - RuntimeError: scheduler_metadata must have... · cc5befbc

Lucas Wilkinson authored Apr 28, 2025


[BugFix] Fix cascade attention - RuntimeError: scheduler_metadata must have shape (metadata_size) (#17283)
Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>

cc5befbc

[BugFix] Fix vllm_flash_attn install issues (#17267) · d8bccde6

Lucas Wilkinson authored Apr 27, 2025


Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
Co-authored-by: Jee Jee Li <pandaleefree@gmail.com>
Co-authored-by: Aaron Pham <contact@aarnphm.xyz>

d8bccde6

27 Apr, 2025 1 commit
- [Bugfix] Get a specific type of layer from forward context (#17222) · 838cedad
  Chen Zhang authored Apr 27, 2025
```
Signed-off-by: Chen Zhang <zhangch99@outlook.com>
```
  838cedad
23 Apr, 2025 1 commit
- [BugFix] llama4 fa3 fix - RuntimeError: scheduler_metadata must have shape (metadata_size) (#16998) · d0da99fb
  Lucas Wilkinson authored Apr 23, 2025
```
Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
```
  d0da99fb
22 Apr, 2025 1 commit

[V1] V1 FlashInfer Attention (#16684) · 986537f1

Michael Goin authored Apr 21, 2025


Signed-off-by: mgoin <mgoin64@gmail.com>
Co-authored-by: Aurick Qiao <qiao@aurick.net>

986537f1

21 Apr, 2025 1 commit
- [TPU][V1] Implicitly adjust page size when there's SMEM OOM (#16871) · 471fe656
  Chengji Yao authored Apr 21, 2025
```
Signed-off-by: Chengji Yao <chengjiyao@google.com>
```
  471fe656
17 Apr, 2025 2 commits
- [Attention] Update to lastest FA3 code (#13111) · 183dad7a
  Lucas Wilkinson authored Apr 17, 2025
```
Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
```
  183dad7a
- [MLA] Simplification to batch P/D reordering (#16673) · 0377b831
  Nick Hill authored Apr 17, 2025
```
Signed-off-by: Nick Hill <nhill@redhat.com>
```
  0377b831
11 Apr, 2025 1 commit
- [Kernel] support merge_attn_states CUDA kernel, 3x speedup (#16173) · e9528f6d
  DefTruth authored Apr 11, 2025
```
Signed-off-by: DefTruth <qiustudent_r@163.com>
```
  e9528f6d
09 Apr, 2025 1 commit
- [BugFix] fix some typos found by typos. (#16314) · 04149cce
  yihong authored Apr 09, 2025
```
Signed-off-by: yihong0618 <zouzou0208@gmail.com>
```
  04149cce
08 Apr, 2025 2 commits
- [BugFix] Fix Llama4 - Index Error When Single Request Near Max Context (#16209) · e1a2c699
  Lucas Wilkinson authored Apr 08, 2025
```
Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
```
  e1a2c699
- Add warning for Attention backends that do not support irope yet (#16212) · 05a015d6
  Yong Hoon Shin authored Apr 07, 2025
  
  05a015d6
07 Apr, 2025 1 commit

Upstream Llama4 Support to Main (#16113) · 55dcce91

Lu Fang authored Apr 07, 2025


Signed-off-by: Aston Zhang <22279212+astonzhang@users.noreply.github.com>
Signed-off-by: Chris Thi <chris.c.thi@gmail.com>
Signed-off-by: drisspg <drisspguessous@gmail.com>
Signed-off-by: Jon Swenson <jmswen@gmail.com>
Signed-off-by: Keyun Tong <tongkeyun@gmail.com>
Signed-off-by: Lu Fang <fanglu@meta.com>
Signed-off-by: Xiaodong Wang <xdwang@meta.com>
Signed-off-by: Yang Chen <yangche@fb.com>
Signed-off-by: Ye (Charlotte) Qi <yeq@meta.com>
Signed-off-by: Yong Hoon Shin <yhshin@meta.com>
Signed-off-by: Zijing Liu <liuzijing2014@gmail.com>
Signed-off-by: Lu Fang <lufang@fb.com>
Signed-off-by: Lu Fang <fanglu@fb.com>
Signed-off-by: Lucia Fang <fanglu@fb.com>
Signed-off-by: Roger Wang <ywang@roblox.com>
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Co-authored-by: Lu Fang <fanglu@fb.com>
Co-authored-by: Roger Wang <ywang@roblox.com>
Co-authored-by: DarkLight1337 <tlleungac@connect.ust.hk>

55dcce91

06 Apr, 2025 1 commit
- [Model] Support Llama4 in vLLM (#16104) · c5752323
  Lu Fang authored Apr 05, 2025
  
  c5752323
04 Apr, 2025 1 commit
- [TPU][V1] Remove ragged attention kernel parameter hard coding (#16041) · fadc59c0
  Chengji Yao authored Apr 04, 2025
```
Signed-off-by: Chengji Yao <chengjiyao@google.com>
```
  fadc59c0
03 Apr, 2025 2 commits

[TPU] Support sliding window and logit soft capping in the paged attention kernel for TPU. (#15732) · b6be6f8d
iefgnoix authored Apr 03, 2025
```
Signed-off-by: Xiongfei Wei <isaacwxf23@gmail.com>
```
b6be6f8d

[ROCM][KERNEL] Paged attention for V1 (#15720) · e73ff24e

Aleksandr Malyshev authored Apr 02, 2025


Signed-off-by: Aleksandr Malyshev <maleksan@amd.com>
Signed-off-by: root <root@banff-cyxtera-s65-4.amd.com>
Co-authored-by: Aleksandr Malyshev <maleksan@amd.com>
Co-authored-by: root <root@banff-cyxtera-s65-4.amd.com>

e73ff24e

29 Mar, 2025 1 commit
- [Kernel][TPU][ragged-paged-attn] vLLM code change for PR#8896 (#15659) · 7c1f7600
  yarongmu-google authored Mar 28, 2025
```
Signed-off-by: Yarong Mu <ymu@google.com>
```
  7c1f7600
23 Mar, 2025 1 commit

[V1] Enable V1 Fp8 cache for FA3 in the oracle (#15191) · dccf535f

Lucas Wilkinson authored Mar 23, 2025


Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>

dccf535f

21 Mar, 2025 3 commits
- [V1] Fix wrong import path of get_flash_attn_version (#15280) · 91ca929d
  Lehua Ding authored Mar 21, 2025
```
Signed-off-by: Lehua Ding <lehuading@tencent.com>
```
  91ca929d
- [V1] Enable Triton(ROCm) Attention backend for Nvidia GPUs (#14071) · f8a08cb9
  Isotr0py authored Mar 21, 2025
```
Signed-off-by: Isotr0py <2037008807@qq.com>
Co-authored-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
```
  f8a08cb9
- [V1] Scheduler Refactoring [1/N] - Add Scheduler Interface (#15250) · 0c6f5023
  Woosuk Kwon authored Mar 20, 2025
```
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
Co-authored-by: Cody Yu <hao.yu.cody@gmail.com>
Co-authored-by: Nick Hill <nhill@redhat.com>
```
  0c6f5023
20 Mar, 2025 1 commit
- [Attention] Flash Attention 3 - fp8 (#14570) · a597a575
  Mickaël Seznec authored Mar 20, 2025
```
Signed-off-by: Mickael Seznec <mickael@mistral.ai>
```
  a597a575
19 Mar, 2025 1 commit
- [V1][TPU] Change kv cache shape. (#15145) · b0e96aae
  iefgnoix authored Mar 19, 2025
```
Signed-off-by: Xiongfei Wei <isaacwxf23@gmail.com>
```
  b0e96aae
15 Mar, 2025 1 commit

[V1] V1 Enablement Oracle (#13726) · d4d93db2

Robert Shaw authored Mar 15, 2025


Signed-off-by: rshaw@neuralmagic.com <rshaw@neuralmagic.com>
Co-authored-by: rshaw@neuralmagic.com <rshaw@neuralmagic.com>
Co-authored-by: Nicolò Lucchesi <nlucches@redhat.com>
Co-authored-by: Tyler Michael Smith <tyler@neuralmagic.com>
Co-authored-by: Michael Goin <michael@neuralmagic.com>

d4d93db2

14 Mar, 2025 1 commit
- [Attention] MLA get rid of materialization (#14770) · 9532c498
  Lucas Wilkinson authored Mar 14, 2025
```
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
```
  9532c498
11 Mar, 2025 1 commit
- dynamic distpatch of fp8 kernels (#14245) · a1c8f379
  Jeff Daily authored Mar 11, 2025
```
Signed-off-by: Jeff Daily <jeff.daily@amd.com>
```
  a1c8f379
10 Mar, 2025 1 commit
- [Perf] Improve MLA on V1 (#14540) · fb0acb6c
  Simon Mo authored Mar 10, 2025
```
Signed-off-by: simon-mo <simon.mo@hey.com>
```
  fb0acb6c
09 Mar, 2025 1 commit
- [V1][TPU] Remove unnecessary padding for running on TPU. (#14467) · 10f75527
  iefgnoix authored Mar 08, 2025
  
  10f75527
08 Mar, 2025 2 commits
- [Bugfix] DeepSeek Accuracy (#14476) · db84f5eb
  Lucas Wilkinson authored Mar 08, 2025
```
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
```
  db84f5eb
- Revert "[Perf] Reduce MLA CPU overheads in V1 (#14384)" (#14471) · ca7a2d5f
  Tyler Michael Smith authored Mar 08, 2025
  
  ca7a2d5f
07 Mar, 2025 3 commits
- Use the optimized block sizes after tuning the kernel. (#14329) · 1e3598ed
  iefgnoix authored Mar 07, 2025
  
  1e3598ed
- [FP8] Refactor apply_fp8_linear and apply_fp8_linear_generic into an object (#14390) · e1744502
  Luka Govedič authored Mar 07, 2025
```
Signed-off-by: luka <luka@neuralmagic.com>
```
  e1744502
- [Perf] Reduce MLA CPU overheads in V1 (#14384) · dae68969
  Lucas Wilkinson authored Mar 06, 2025
```
Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
```
  dae68969
06 Mar, 2025 2 commits
- [Hardware][TPU]Enable ragged paged attention kernel and resolve recompilation issue (#14310) · 0578e5a4
  Chengji Yao authored Mar 06, 2025
```
Signed-off-by: Chengji Yao <chengjiyao@google.com>
```
  0578e5a4
- [V1][Bugfix] Standardize quantized kv cache rejection for attention backends (#14221) · 6832707e
  Michael Goin authored Mar 06, 2025
```
Signed-off-by: mgoin <mgoin64@gmail.com>
```
  6832707e