Commits · aa1356ec53a65a79a0027e8c265c76d84de8c046 · OpenDAS / vllm_cscc

"deployment/vscode:/vscode.git/clone" did not exist on "bcefce6a814c2c1228e25fcd98c88fea9e8111a9"

21 Oct, 2025 11 commits

[ROCM] Enable CompressedTensorsWNA16 (#27187) · ba09652d
JartX authored Oct 21, 2025
```
Signed-off-by: JartX <sagformas@epdcenter.es>
```
ba09652d
[Deepseek v3.2] Optimize top_k_per_row (#26763) · 80e94529
Daniel Cámpora authored Oct 21, 2025
```
Signed-off-by: Daniel Campora <961215+dcampora@users.noreply.github.com>
```
80e94529
[MM][Core] Decouple ViT backend from LM backend (#27061) · c3a2c6ac
Roger Wang authored Oct 21, 2025
```
Signed-off-by: Roger Wang <hey@rogerw.io>
```
c3a2c6ac
[Fix][Spec Decode] Fix llama4 draft loading with different quantization (#27136) · be444507
Zebing Lin authored Oct 21, 2025
```
Signed-off-by: linzebing <linzebing1995@gmail.com>
```
be444507
[Bugfix] Fix broken MTP weight loading for FP8 KV Scales (#27227) · f381cf23
Benjamin Chislett authored Oct 21, 2025
```
Signed-off-by: Benjamin Chislett <bchislett@nvidia.com>
```
f381cf23

[Bugfix] Fix gpt-oss w4a8 DP/EP on B200 (#26729) · 5ff5d94e

Varun Sundar Rabindranath authored Oct 21, 2025


Signed-off-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
Co-authored-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
Co-authored-by: Michael Goin <mgoin64@gmail.com>

5ff5d94e

[ModelOpt] Load w13/w2_input_scale for all experts, nvfp4 (#26135) · f95da13c

Shu Wang authored Oct 21, 2025


Signed-off-by: Shu Wang <shuw@nvidia.com>
Signed-off-by: Shu Wang. <shuw@nvidia.com>
Co-authored-by: Michael Goin <mgoin64@gmail.com>

f95da13c

[BugFix] GPT-OSS Attention DP + MoE TP weight loading issue (#24032) · aef368aa
Po-Han Huang (NVIDIA) authored Oct 21, 2025
```
Signed-off-by: Po-Han Huang <pohanh@nvidia.com>
```
aef368aa

[Feature][Kernel]FusedMoE LoRA (#21229) · 5f6cbf60

Chen Wu authored Oct 21, 2025


Signed-off-by: wuchen <cntryroa@gmail.com>
Signed-off-by: banjuede <lmklhc@163.com>
Signed-off-by: Chen Wu <cntryroa@gmail.com>
Signed-off-by: Danielle Robinson <dmmaddix@amazon.com>
Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
Signed-off-by: bk-201 <joy25810@foxmail.com>
Co-authored-by: wuchen <wuchen@zetyun.com>
Co-authored-by: Nathan Van Gheem <vangheem@gmail.com>
Co-authored-by: banjuede <lmklhc@163.com>
Co-authored-by: Danielle Robinson <dmmaddix@amazon.com>
Co-authored-by: Jee Jee Li <pandaleefree@gmail.com>
Co-authored-by: bk-201 <joy25810@foxmail.com>

5f6cbf60

[cpu] Dispatch un-quantized linear to oneDNN/ACL by default for AArch64 (#27183) · 163965d1
Fadi Arafeh authored Oct 21, 2025
```
Signed-off-by: Fadi Arafeh <fadi.arafeh@arm.com>
Co-authored-by: Michael Yang <Michael.Yang@arm.com>
```
163965d1
[Quantization] Automatically infer AWQ `modules_to_not_convert` field (#26909) · 352c0c8a
Isotr0py authored Oct 21, 2025
```
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
```
352c0c8a

20 Oct, 2025 5 commits

[Feature][Quantization] auto_round support for mixed bits quantization (#23812) · 87778d5f

Heng Guo authored Oct 21, 2025

Signed-off-by: n1ck-guo <heng.guo@intel.com>
Signed-off-by: Heng Guo <heng.guo@intel.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>

87778d5f

[Kernel][Model] Tune fused_moe Triton configs for Qwen3-30B A3/A3B on H100 (FP8/BF16) (#26268) · 4d0f2661
shivampr authored Oct 20, 2025
```
Signed-off-by: Shivam <shivampr.dev@gmail.com>
```
4d0f2661

Nemotron Nano V2 VL + EVS Video Support (#27107) · e93ff6c8

Eugene Khvedchenya authored Oct 20, 2025


Signed-off-by: Eugene Khvedchenia <ekhvedchenia@nvidia.com>
Signed-off-by: Natan Bagrov <nbagrov@nvidia.com>
Signed-off-by: Roger Wang <hey@rogerw.io>
Co-authored-by: Natan Bagrov <nbagrov@nvidia.com>
Co-authored-by: Roger Wang <hey@rogerw.io>

e93ff6c8

[Kernel] Accelerate solve_tril with TMA (#26746) · 9fce7bee
Jiangyun Zhu authored Oct 20, 2025
```
Signed-off-by: zjy0516 <riverclouds.zhu@qq.com>
```
9fce7bee

[Model][VLM] Support Bee-8B Model (#27012) · f32bf758

Yi Zhang authored Oct 20, 2025


Signed-off-by: uyzhang <yi.zhang.4096@gmail.com>
Signed-off-by: Yi Zhang <zhangyi970819@gmail.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Co-authored-by: Roger Wang <hey@rogerw.io>

f32bf758

19 Oct, 2025 2 commits
- [Misc] Move utils to avoid conflicts with stdlib, and move tests (#27169) · d31f7844
  Cyrus Leung authored Oct 19, 2025
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  d31f7844
- output type conversion fix (#27159) · 221bf725
  Jianyu Huang authored Oct 19, 2025
  
  221bf725
18 Oct, 2025 7 commits
- [BugFix] Disable fp8 kv-cache by default for DeepSeek V3.2 (#27121) · c2bba690
  Lucas Wilkinson authored Oct 18, 2025
```
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
Signed-off-by: Lucas Wilkinson <LucasWilkinson@users.noreply.github.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
```
  c2bba690
- [Chore] Clean up pytorch helper functions in `vllm.utils` (#26908) · 6ac5e06f
  Isotr0py authored Oct 19, 2025
```
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
Signed-off-by: isotr0py <2037008807@qq.com>
```
  6ac5e06f
- [Models][QwenVL] Remove unnecessary `.contiguous()` calls (#27106) · 5c2acb27
  Lukas Geiger authored Oct 18, 2025
```
Signed-off-by: Lukas Geiger <lukas.geiger94@gmail.com>
```
  5c2acb27
- [Misc] Refactor `get_kv_cache_spec` into `AttentionLayerBase` (#26587) · b26b70be
  Nicolò Lucchesi authored Oct 18, 2025
```
Signed-off-by: NickLucche <nlucches@redhat.com>
```
  b26b70be
- [Feature] Batch Invariant: Support DeepGEMM and Blackwell (#27127) · 245e4f2c
  Wentao Ye authored Oct 18, 2025
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  245e4f2c
- [Misc] Rev DeepEP (#27122) · 30a33b92
  Varun Sundar Rabindranath authored Oct 18, 2025
```
Signed-off-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
Co-authored-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
```
  30a33b92
- [Perf] Add H100 fused MoE config (#25398) · c981f0ea
  ZiTian Zhao authored Oct 18, 2025
```
Signed-off-by: zitian.zhao <zitian.zhao@tencentmusic.com>
```
  c981f0ea
17 Oct, 2025 15 commits
- [Minor] Remove unnecessary error message (#27115) · d29483b5
  Zhuohan Li authored Oct 17, 2025
```
Signed-off-by: Zhuohan Li <zhuohan123@gmail.com>
```
  d29483b5
- [Chore] Remove unused `PolyNorm` layer (#27110) · 3125d799
  Isotr0py authored Oct 18, 2025
```
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
```
  3125d799
- [Bugfix] [AITER] [ROCm] Fix Quark MoE Quant Config and AITER Fused MoE quant type logic (#27029) · e33ee23e
  vllmellm authored Oct 18, 2025
```
Signed-off-by: vllmellm <vllm.ellm@embeddedllm.com>
```
  e33ee23e
- [ROCM] MoE fp4 CK kernel (#26545) · 0925b28a
  Aleksandr Malyshev authored Oct 17, 2025
```
Signed-off-by: Aleksandr Malyshev <maleksan@amd.com>
Co-authored-by: Aleksandr Malyshev <maleksan@amd.com>
```
  0925b28a
- [bugfix] Qwen3-VL fix video incorrect timestamp calculations while do_sample_frames=True (#27104) · 4c91a28e
  燃 authored Oct 18, 2025
```
Co-authored-by: 松灵 <wpf272043@alibaba-inc.com>
```
  4c91a28e
- [torch.compile] Enable attention and allreduce fusion without custom ops enabled (#24604) · bd7157a0
  Luka Govedič authored Oct 17, 2025
```
Signed-off-by: Luka Govedič <lgovedic@redhat.com>
Signed-off-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
```
  bd7157a0
- [Harware][AMD][Model] Triton MoE tuning configs for GLM-4.5 for MI350 and MI355 (#25586) · c253745e
  Reima Karhila (AMD) authored Oct 17, 2025
```
Signed-off-by: Reima Karhila <reima.karhila@amd.com>
Signed-off-by: xaguilar <Xavier.AguilarFruto@amd.com>
Co-authored-by: xaguilar <Xavier.AguilarFruto@amd.com>
```
  c253745e
- [Model]Improve Qwen3VLMoeForConditionalGeneration packed_modules_mapping (#27096) · daec4d26
  Jee Jee Li authored Oct 17, 2025
```
Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
```
  daec4d26
- [Docs] Replace `rst` style double-backtick with `md` single-backtick (#27091) · 6c9fdbf7
  Harry Mellor authored Oct 17, 2025
```
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
```
  6c9fdbf7
- [VLM][Refactor] Remove useless func `get_input_positions` in `MRotaryEmbedding` (#27088) · e20eba75
  Mengqing Cao authored Oct 17, 2025
```
Signed-off-by: MengqingCao <cmq0113@163.com>
```
  e20eba75
- [Kernel][Performance] Fuse float cast and renormalize to topk softmax kernel (#26717) · 75c7ad99
  zhrrr authored Oct 17, 2025
```
Signed-off-by: zhuhaoran <zhuhaoran.zhr@alibaba-inc.com>
Signed-off-by: izhuhaoran <izhuhaoran@qq.com>
```
  75c7ad99
- [Model] Add support for LightOnOCR (#26916) · 3aeb19a3
  Said Taghadouini authored Oct 17, 2025
```
Signed-off-by: Said Taghadouini <taghadouinisaid@gmail.com>
Signed-off-by: Said Taghadouini <84044788+staghado@users.noreply.github.com>
Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>
```
  3aeb19a3
- [Model] Always use Transformers backend for PaliGemma and Gemma3-MM (#26715) · 8c017b34
  Cyrus Leung authored Oct 17, 2025
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  8c017b34
- [Bugfix][Qwen] fixes the weights dtype in qwen3_next: it is actually a bfloat16 (#27030) · bde9e227
  Tao He authored Oct 17, 2025
```
Signed-off-by: Tao He <linzhu.ht@alibaba-inc.com>
```
  bde9e227
- disable graph partition in custom op (#26952) · 08405609
  Boyuan Feng authored Oct 16, 2025
```
Signed-off-by: Boyuan Feng <boyuan@meta.com>
Signed-off-by: Boyuan Feng <fby.1994@gmail.com>
Co-authored-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
```
  08405609