Commits · 1983609239caaab24ab1ed2bfa2aa92e8c76c1b1 · OpenDAS / vllm_cscc

24 Sep, 2025 2 commits

[Bug] Fix AttributeError: 'FusedMoE' object has no attribute... · 88d7bdbd

Wentao Ye authored Sep 23, 2025


[Bug] Fix AttributeError: 'FusedMoE' object has no attribute 'w13_weight_scale'. Did you mean: 'w13_weight_scale_inv' (#25519)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
Signed-off-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>

88d7bdbd

Add CUTLASS FP8 MOE benchmark scripts and kernel config (#25302) · 0d235b87
Chenxi Yang authored Sep 23, 2025
```
Signed-off-by: Chenxi Yang <cxyang@fb.com>
Co-authored-by: Chenxi Yang <cxyang@fb.com>
```
0d235b87

23 Sep, 2025 19 commits

[ROCm] Small functional changes for gptoss (#25201) · bde2a1a8
Juan Villamizar authored Sep 23, 2025
```
Signed-off-by: jpvillam <jpvillam@amd.com>
Co-authored-by: jpvillam <jpvillam@amd.com>
```
bde2a1a8

[Kernel] [Mamba] Remove BLOCK_H=1 from list of tuneable configurations for... · 5e25b122

Thomas Parnell authored Sep 24, 2025


[Kernel] [Mamba] Remove BLOCK_H=1 from list of tuneable configurations for `_chunk_cumsum_fwd_kernel` (#25197)
Signed-off-by: Thomas Parnell <tpa@zurich.ibm.com>
Co-authored-by: Chih-Chieh-Yang <chih.chieh.yang@ibm.com>

5e25b122

Remove redundant mutates_args and dispatch_key for direct_register_custom_op (#25512) · 7361ab37
Michael Goin authored Sep 23, 2025
```
Signed-off-by: mgoin <mgoin64@gmail.com>
```
7361ab37
[CI/Build] Fix and re-enable v1 PP test on CI (#25496) · ae002924
Isotr0py authored Sep 24, 2025
```
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
```
ae002924
[Bugfix] Fix for the import error from #24588 (#25481) · 690f948e
Gregory Shtrasberg authored Sep 23, 2025
```
Signed-off-by: Gregory Shtrasberg <Gregory.Shtrasberg@amd.com>
```
690f948e
[Bugfix] Lower gpt-oss max cudagraph size to 992 to be compatible with FA3 (#25508) · a8ffc4f0
Michael Goin authored Sep 23, 2025
```
Signed-off-by: mgoin <mgoin64@gmail.com>
```
a8ffc4f0
[Speculators][Speculative Decoding] Fix gpt-oss eagle3 accuracy issue (#25406) · d5944d51
jiahanc authored Sep 23, 2025
```
Signed-off-by: jiahanc <173873397+jiahanc@users.noreply.github.com>
```
d5944d51

[Performance] Move apply_w8a8_block_fp8_linear to an op class (#24666) · 63400259

ElizaWszola authored Sep 23, 2025


Signed-off-by: ElizaWszola <ewszola@redhat.com>
Signed-off-by: ElizaWszola <elizaw.9289@gmail.com>
Signed-off-by: Luka Govedič <lgovedic@redhat.com>
Signed-off-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
Co-authored-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
Co-authored-by: Michael Goin <mgoin64@gmail.com>
Co-authored-by: Luka Govedič <lgovedic@redhat.com>

63400259

[ROCm] Add skinny gemm bias support for dtypes fp16,bf16,fp8 (#24988) · a3a78280

Hashem Hashemi authored Sep 23, 2025


Signed-off-by: Hashem Hashemi <hashem.hashemi@amd.com>
Signed-off-by: Hashem Hashemi <159079214+amd-hhashemi@users.noreply.github.com>

a3a78280

[Kernels] Support blocked fp8 quantization for compressed tensors MoE (#25219) · f11e3c51
bnellnm authored Sep 23, 2025
```
Signed-off-by: Bill Nell <bnell@redhat.com>
Co-authored-by: Michael Goin <mgoin64@gmail.com>
```
f11e3c51

[Core/DBO][2/N] Dual-Batch Overlap add DeepEP High Throughput support and Prefill support (#24845) · cc1dc7ed

Lucas Wilkinson authored Sep 23, 2025


Signed-off-by: Sage Moore <sage@neuralmagic.com>
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
Signed-off-by: yewentao256 <zhyanwentao@126.com>
Signed-off-by: Lucas Wilkinson <LucasWilkinson@users.noreply.github.com>
Signed-off-by: Tyler Michael Smith <tyler@neuralmagic.com>
Co-authored-by: Sage Moore <sage@neuralmagic.com>
Co-authored-by: yewentao256 <zhyanwentao@126.com>
Co-authored-by: Tyler Michael Smith <tyler@neuralmagic.com>

cc1dc7ed

[V1] Remove V0 code paths for Hybrid models (#25400) · a903669e
Thomas Parnell authored Sep 23, 2025
```
Signed-off-by: Thomas Parnell <tpa@zurich.ibm.com>
```
a903669e

[BugFix] Register expert_map as named buffer for wake_up and sleep (#25458) · 61d1b355

Joel authored Sep 23, 2025


Signed-off-by: wuxibin <wuxibin@bytedance.com>
Signed-off-by: youkaichao <youkaichao@gmail.com>
Co-authored-by: youkaichao <youkaichao@gmail.com>

61d1b355

[Bugfix] Fix idefics3 `tie_word_embeddings` (#25454) · 231c2c63
Isotr0py authored Sep 23, 2025
```
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
```
231c2c63
[Misc] Move DP for ViT code inside model executor dir (#25459) · babad6e5
Cyrus Leung authored Sep 23, 2025
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
babad6e5
Handle triton kernel import exception (#25319) · ba8d2165
Ming Yang authored Sep 23, 2025
```
Signed-off-by: Ming Yang <minos.future@gmail.com>
```
ba8d2165
[Model] Enable DP for ViT in Qwen2-VL (#25445) · c98be0a2
Cyrus Leung authored Sep 23, 2025
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
c98be0a2
[DP/EP][GPTOSS] Use triton matmul-ogs kernels for GPTOSS DP/EP (#24588) · e8db44f8
Varun Sundar Rabindranath authored Sep 23, 2025
```
Signed-off-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
Co-authored-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
```
e8db44f8

[feat] Support MRoPE + YaRN (#25384) · fc97733d

JJJYmmm authored Sep 23, 2025


Signed-off-by: liuye.hj <liuye.hj@alibaba-inc.com>
Co-authored-by: liuye.hj <liuye.hj@alibaba-inc.com>

fc97733d

22 Sep, 2025 5 commits

[EPLB] Reduce EPLB Inference Overhead (#24573) · 06a41334

Bowen Wang authored Sep 22, 2025

Signed-off-by: Bowen Wang <abmfy@icloud.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Co-authored-by: Tyler Michael Smith <tyler@neuralmagic.com>

06a41334

[Kernel] MI-300X triton moe configs (#23445) · ac243886
Sara-KS authored Sep 22, 2025
```
Signed-off-by: Sara Kokkila Schumacher <saraks@ibm.com>
```
ac243886
[TPU] Deprecate `xm.mark_step` in favor of ``torch_xla.sync` (#25254) · 4cf71cc8
Nicolò Lucchesi authored Sep 22, 2025
```
Signed-off-by: NickLucche <nlucches@redhat.com>
Co-authored-by: Ye (Charlotte) Qi <yeq@meta.com>
```
4cf71cc8
Enable Eagle3 speculative decoding for GPT-OSS model (#25246) · 21467f9a
Eldar Kurtić authored Sep 22, 2025
```
Signed-off-by: Eldar Kurtic <8884008+eldarkurtic@users.noreply.github.com>
```
21467f9a

[Model] Support Dots OCR (#24645) · 7b57a433

Roger Wang authored Sep 21, 2025


Signed-off-by: Roger Wang <hey@rogerw.io>
Co-authored-by: yinz-aizip <yinz@aizip.ai>

7b57a433

21 Sep, 2025 8 commits
- Remove V0 attention backends (#25351) · bc6e542d
  Woosuk Kwon authored Sep 21, 2025
```
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
```
  bc6e542d
- [Perf] Further optimization for Qwen3-VL `fast_pos_embed_interpolate` (#25347) · af7dfb0d
  Isotr0py authored Sep 22, 2025
```
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
```
  af7dfb0d
- [V0 Deprecation] Remove V0 sampling metadata (#25345) · 1c3ffdbe
  Woosuk Kwon authored Sep 21, 2025
```
Signed-off-by: Woosuk Kwon <woosuk@thinkingmachines.ai>
```
  1c3ffdbe
- [V0 Deprecation] Remove V0 Sequence class & Sampler (#25332) · 26e673fe
  Woosuk Kwon authored Sep 21, 2025
```
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
Signed-off-by: Woosuk Kwon <woosuk@thinkingmachines.ai>
```
  26e673fe
- [MM][Perf] Minor Optimization on Qwen3-VL `fast_pos_embed_interpolate` (#25337) · 30d08911
  Roger Wang authored Sep 21, 2025
```
Signed-off-by: Roger Wang <hey@rogerw.io>
```
  30d08911
- [Multi Modal][Performance] Fused Q,K's apply_rope in more models (#25005) · 035fd2bd
  Wenlong Wang authored Sep 20, 2025
```
Signed-off-by: wwl2755 <wangwenlong2755@gmail.com>
Co-authored-by: Roger Wang <hey@rogerw.io>
```
  035fd2bd
- [Chore] Remove unused sampler in models (#25324) · 572ddf83
  Woosuk Kwon authored Sep 20, 2025
```
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
```
  572ddf83
- [V0 Deprecation] Remove LLMEngine (#25033) · 52c2a8d4
  Woosuk Kwon authored Sep 20, 2025
```
Signed-off-by: Woosuk Kwon <woosuk@thinkingmachines.ai>
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
```
  52c2a8d4
20 Sep, 2025 6 commits
- [Model] Cleanup InternViT's data parallel implementation (#25306) · 3c713a97
  Isotr0py authored Sep 20, 2025
```
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
```
  3c713a97
- Generate _ModelInfo properties file when loading to improve loading speed (#23558) · bf8b26ca
  Manoel Marques authored Sep 20, 2025
```
Signed-off-by: Manoel Marques <manoel.marques@ibm.com>
Signed-off-by: Manoel Marques <manoelmrqs@gmail.com>
Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
```
  bf8b26ca
- [CI Failure] Disable FlashInfer RoPE to unblock CI (#25299) · e08a3a3f
  Michael Goin authored Sep 20, 2025
```
Signed-off-by: mgoin <mgoin64@gmail.com>
```
  e08a3a3f
- [Bugfix] Fix Qwen3-VL-MoE weight loading for EP (#25300) · be874c02
  Roger Wang authored Sep 20, 2025
```
Signed-off-by: Roger Wang <hey@rogerw.io>
```
  be874c02
- [Optimization] Avoid repeated model architecture conversion for pooling models (#25261) · c60e6137
  Cyrus Leung authored Sep 20, 2025
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  c60e6137
- [BUGFIX] GPTQ quantization compatibility for Qwen3 Next MOE models (AutoGPTQ... · 36429096
  JartX authored Sep 20, 2025
```
[BUGFIX] GPTQ quantization compatibility for Qwen3 Next MOE models (AutoGPTQ and AutoRound-GPTQ) (#25268)
Signed-off-by: JartX <sagformas@epdcenter.es>
```
  36429096