Commits · 2aab2bb54366c5a26add1b07f107b86f7fe28ff5 · OpenDAS / vllm_cscc

"lib/llm/vscode:/vscode.git/clone" did not exist on "51b4cd7e880d7bd7a1bbec344eadbe07e7546a14"

21 Feb, 2026 1 commit

[ROCm][Bugfix]: Only save unpadded sizes for shared_experts in MoERunner to... · ded333fb

Rohan Potdar authored Feb 20, 2026


[ROCm][Bugfix]: Only save unpadded sizes for shared_experts in MoERunner to fix rmsnorm pad fusion (#34636)
Signed-off-by: Rohan138 <rohanpotdar138@gmail.com>

ded333fb

20 Feb, 2026 4 commits

[CI] Revert PRs 34818 and 33600 (#34979) · aaefc58e
Lucas Wilkinson authored Feb 20, 2026

aaefc58e

[Minor] Add logging when using MXFP4 MXFP8 TRTLLM backend (#34916) · 05977923

Frank Wang authored Feb 19, 2026

Signed-off-by: frankwang28 <frank.wbb@hotmail.com>
Signed-off-by: Frank Wang <41319051+frankwang28@users.noreply.github.com>
Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>

05977923

[Model Bash]: Improve FP8 Oracle for Config Specific Kernel Selection (#34260) · 81bfc21a

Elizabeth Thomas authored Feb 19, 2026

Signed-off-by: Elizabeth Thomas <email2eliza@gmail.com>
Signed-off-by: Robert Shaw <robertgshaw2-redhat@h100-02.nemg-001.lab.rdu2.dc.redhat.com>
Signed-off-by: Robert Shaw <robertgshaw2@gmail.com>
Co-authored-by: Robert Shaw <robertgshaw2-redhat@h100-02.nemg-001.lab.rdu2.dc.redhat.com>
Co-authored-by: Robert Shaw <robertgshaw2@gmail.com>

81bfc21a

[Quark] Fix MoE fp8 activation scale handling on mi300 (#34386) · d9e62c03
Bowen Bao authored Feb 19, 2026
```
Signed-off-by: Bowen Bao <bowenbao@amd.com>
```
d9e62c03

19 Feb, 2026 6 commits
- [UX] More descriptive reasons in is_supported_config for MoE (#34908) · f72061a1
  Michael Goin authored Feb 19, 2026
```
Signed-off-by: mgoin <mgoin64@gmail.com>
```
  f72061a1
- [Bugfix] Fix Basic Models Test (#34818) · 662205d3
  Matthew Bonanni authored Feb 19, 2026
```
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
Co-authored-by: Robert Shaw <114415538+robertgshaw2-redhat@users.noreply.github.com>
```
  662205d3
- [Bugfix] Fix cutlass fp8 kernel on hopper for Qwen3.5 (#34914) · 4fb8beef
  Roger Wang authored Feb 19, 2026
```
Signed-off-by: Roger Wang <hey@rogerw.io>
```
  4fb8beef
- [Refactor] Deprecate `head_first` for `chunk_gated_delta_rule` (#34263) · c683d11c
  Wentao Ye authored Feb 19, 2026
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  c683d11c
- [Model Bash][DeepSeekR1] Remove Shared Expert Clone (#34344) · 4685a630
  Robert Shaw authored Feb 19, 2026
```
Signed-off-by: Robert Shaw <robshaw@redhat.com>
Co-authored-by: Robert Shaw <robshaw@redhat.com>
```
  4685a630
- fix(docs): fix typos in comments and docstrings (#34836) · ad5aa6bd
  Manrique Vargas authored Feb 19, 2026
```
Signed-off-by: machov <mv1742@nyu.edu>
```
  ad5aa6bd
18 Feb, 2026 9 commits
- [MoE Refactor] Convert mxfp4 marlin into modular kernel format (#34588) · 40da9625
  Yongye Zhu authored Feb 18, 2026
```
Signed-off-by: Yongye Zhu <zyy1102000@gmail.com>
Co-authored-by: Robert Shaw <114415538+robertgshaw2-redhat@users.noreply.github.com>
```
  40da9625
- [Bugfix][MoE Kernel] Fix incorrect routing selection for models without expert... · 847a57cd
  Wenlong Wang authored Feb 18, 2026
```
[Bugfix][MoE Kernel] Fix incorrect routing selection for models without expert groups (e.g., MiniMax-M2.1) (#34673)
Signed-off-by: wwl2755 <wangwenlong2755@gmail.com>
Signed-off-by: Robert Shaw <robshaw@redhat.com>
Co-authored-by: Robert Shaw <robshaw@redhat.com>
Co-authored-by: Robert Shaw <114415538+robertgshaw2-redhat@users.noreply.github.com>
```
  847a57cd
- [Bugfix] Remove assert causing hipErrorStreamCaptureUnsupported (#34455) · 0e60c925
  Jaden Mathias authored Feb 18, 2026
```
Signed-off-by: Jaden Mathias <jaden.mathias@amd.com>
```
  0e60c925
- [Bugfix] Redo Qwen3.5/Qwen3-Next GDN projector fusion (#34697) · c0bd8b13
  Isotr0py authored Feb 19, 2026
```
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
Co-authored-by: JJJYmmm <92386084+JJJYmmm@users.noreply.github.com>
```
  c0bd8b13
- [Bugfix] Fix NVFP4 TRTLLM MoE non-gated support; add gsm8k for Nemotron-3-Nano FP8+NVFP4 (#34725) · caeb887b
  Michael Goin authored Feb 18, 2026
```
Signed-off-by: mgoin <mgoin64@gmail.com>
```
  caeb887b
- [Model Bash] DeepSeek R1 BF16 Min Latency QKV A GEMM (0.5% E2E Speedup) (#34758) · 6874638b
  Robert Shaw authored Feb 18, 2026
```
Signed-off-by: Robert Shaw <robshaw@redhat.com>
Co-authored-by: Robert Shaw <robshaw@redhat.com>
```
  6874638b
- [Quantization] - Added uses_meta_device_weights to quant config (#34645) · 1faa8cb7
  Asaf Joseph Gardin authored Feb 18, 2026
```
Signed-off-by: Josephasafg <ajgard7@gmail.com>
```
  1faa8cb7
- [Bugfix] fix activation in cpu_fused_moe_torch call (#34696) · e89a91d9
  Marek Michalowski authored Feb 18, 2026
```
Signed-off-by: Marek Michalowski <marek.michalowski@arm.com>
```
  e89a91d9
- [Bugfix] Fix quant RMS norm fusion for quantization with TMA-aligned scales (#33255) · a88b3be7
  ElizaWszola authored Feb 18, 2026
```
Signed-off-by: ElizaWszola <ewszola@redhat.com>
Co-authored-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
```
  a88b3be7
17 Feb, 2026 3 commits
- Remove dead bitsandbytes CxB code from 8-bit inference path (#34633) · d44a5b6c
  Tim Dettmers authored Feb 17, 2026
```
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
```
  d44a5b6c
- Revert "[Models] Fuse Qwen3.5 GDN's qkvz_proj and ba_proj" (#34683) · 1d65283e
  Jiangyun Zhu authored Feb 17, 2026
  
  1d65283e
- [Bugfix] Fix fused MoE int32 overflow in stride*offset without perf regression (#34507) · b68fd899
  haosdent authored Feb 17, 2026
```
Signed-off-by: haosdent <haosdent@gmail.com>
Co-authored-by: Michael Goin <mgoin64@gmail.com>
```
  b68fd899
16 Feb, 2026 5 commits
- [NemotronH] Do not force router to run in fp32 (#34582) · 3b30e615
  roikoren755 authored Feb 16, 2026
```
Signed-off-by: Roi Koren <roik@nvidia.com>
```
  3b30e615
- Fix call to moe_mk in modelopt MoE modules (required for LoRA) (#34575) · ec7d9e67
  danisereb authored Feb 16, 2026
```
Signed-off-by: Daniel Serebrenik <daserebrenik@nvidia.com>
```
  ec7d9e67
- [Models] Fuse Qwen3.5 GDN's qkvz_proj and ba_proj (#34492) · 3bb4e431
  Isotr0py authored Feb 16, 2026
```
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
```
  3bb4e431
- [Bugfix] Fix Dynamo unexpected keyword argument (#34320) · a5ccc85c
  Samu Tamminen authored Feb 16, 2026
```
Signed-off-by: Samu Tamminen <stammine@amd.com>
Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
```
  a5ccc85c
- [Bugfix] Add method to swap quant_method on FusedMoE to fix LoRA issues (#34453) · 5bff999d
  bnellnm authored Feb 15, 2026
```
Signed-off-by: Bill Nell <bnell@redhat.com>
```
  5bff999d
15 Feb, 2026 3 commits
- [MM Encoder] Add Triton ViT attention backend (#32183) · 71cd8926
  Isotr0py authored Feb 15, 2026
```
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
```
  71cd8926
- [Bugfix] Handle num_expert_group=None in flashinfer block-scale FP8 MoE (#34494) · 79f3fab0
  haosdent authored Feb 15, 2026
```
Signed-off-by: haosdent <haosdent@gmail.com>
```
  79f3fab0
- [BUGFIX] Fix accuracy regression for NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4 with TP>1 (#34476) · 604b9eae
  Vadim Gimpelson authored Feb 15, 2026
```
Signed-off-by: Vadim Gimpelson <vadim.gimpelson@gmail.com>
```
  604b9eae
13 Feb, 2026 4 commits
- Revert "[Bugfix] Fix fused MoE IMA (sans chunking) by using int64 for strides" (#34530) · bfaa5593
  Michael Goin authored Feb 13, 2026
  
  bfaa5593
- [Perf] fused_moe: add int4_w4a16 benchmark support and tuning config (#34130) · 934acdde
  Matthias Gehre authored Feb 13, 2026
```
Signed-off-by: Matthias Gehre <matthias.gehre@amd.com>
Co-authored-by: TJian <tunjian.tan@embeddedllm.com>
```
  934acdde
- [Bugfix] Fix mamba state dtype setting for Qwen3-Next and Qwen3.5 (#34489) · eea3024f
  Roger Wang authored Feb 12, 2026
```
Signed-off-by: Roger Wang <hey@rogerw.io>
```
  eea3024f
- [Bugfix] Remove assert that's no longer valid (#34443) · 04ea31ba
  bnellnm authored Feb 12, 2026
```
Signed-off-by: Bill Nell <bnell@redhat.com>
```
  04ea31ba
12 Feb, 2026 5 commits
- [Kernel] Support Flashinfer trtllm fused MoE non gated FP8 & NVFP4 (#33506) · f120bd42
  amitz-nv authored Feb 12, 2026
```
Signed-off-by: amitz-nv <203509407+amitz-nv@users.noreply.github.com>
```
  f120bd42
- [Bugfix] Enforce DeepGEMM when using sparse_attn_indexer on CUDA (#34374) · 6d4e27ce
  Michael Goin authored Feb 12, 2026
```
Signed-off-by: mgoin <mgoin64@gmail.com>
```
  6d4e27ce
- [ROCm][quantization] improve OCP weight quant parser robust (#34431) · 766e1678
  xuebwang-amd authored Feb 13, 2026
```
Signed-off-by: xuebwang-amd <xuebwang@amd.com>
Co-authored-by: TJian <tunjian.tan@embeddedllm.com>
```
  766e1678
- [Attention] Add FlashInfer Sparse MLA backend (#33451) · f2c47886
  Matthew Bonanni authored Feb 12, 2026
```
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
Co-authored-by: Lucas Wilkinson <lwilkins@redhat.com>
Co-authored-by: Lucas Wilkinson <LucasWilkinson@users.noreply.github.com>
```
  f2c47886
- Add config file for fused MoE for Nemotron (TP4, B200) (#34411) · dea63512
  danisereb authored Feb 12, 2026
```
Signed-off-by: Daniel Serebrenik <daserebrenik@nvidia.com>
```
  dea63512