Commits · 5234dc74514a6b3d0740b39f56a4a4208ec86ecc · OpenDAS / vllm_cscc · GitLab

01 Oct, 2025 17 commits
- [NVIDIA] Blackwell Family (#24673) · 5234dc74
  Johnny authored Oct 01, 2025
```
Signed-off-by: Johnny <johnnynuca14@gmail.com>
Signed-off-by: johnnynunez <johnnynuca14@gmail.com>
Signed-off-by: Johnny <johnnync13@gmail.com>
Signed-off-by: Salvatore Cena <cena@cenas.it>
Co-authored-by: Aidyn-A <31858918+Aidyn-A@users.noreply.github.com>
Co-authored-by: Salvatore Cena <cena@cenas.it>
```
  5234dc74
- [Bugfix] Apply same sampling parameters for both `n=1` and `n>1` (#26005) · 3b7c20a6
  Kenichi Maehashi authored Oct 01, 2025
```
Signed-off-by: Kenichi Maehashi <maehashi@preferred.jp>
```
  3b7c20a6
- [Benchmark] Finish documented v0.11.0 deprecation of --endpoint-type (#26007) · f9e71481
  Nathan Scott authored Oct 01, 2025
```
Signed-off-by: Nathan Scott <nathans@redhat.com>
```
  f9e71481
- [MISC] Fix misleading batch_size_capture_list when cuda_graph_sizes < 4 (#25829) · 2518230d
  billishyahao authored Oct 01, 2025
```
Signed-off-by: billishyahao <bill.he@amd.com>
Co-authored-by: Luka Govedic <ProExpertProg@users.noreply.github.com>
```
  2518230d
- [CI] Only capture a single CUDA graph size in CI by default (#25951) · a332b845
  Harry Mellor authored Oct 01, 2025
```
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
```
  a332b845
- [Misc] Factor out common `_apply_feature_select_strategy` (#26003) · 1405f0c7
  Cyrus Leung authored Oct 01, 2025
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  1405f0c7
- [BugFix][MM] Fix Nonetype error when video is cache in qwen2.5-omni-thinker (#26004) · 84d57342
  Wenlong Wang authored Oct 01, 2025
```
Signed-off-by: wwl2755 <wangwenlong2755@gmail.com>
```
  84d57342
- [Doc] updating torch.compile doc link (#25989) · 57b46d76
  nadathurv authored Oct 01, 2025
```
Signed-off-by: nadathurv <work.vnadathur@gmail.com>
Signed-off-by: WorldExplored <srreyansh.sethi@gmail.com>
Co-authored-by: Srreyansh Sethi <107075589+WorldExplored@users.noreply.github.com>
```
  57b46d76
- [Misc]allow disable pynccl (#25421) · f48b6a03
  Lucia Fang authored Sep 30, 2025
```
Signed-off-by: Lu Fang <fanglu@fb.com>
Co-authored-by: Lucia (Lu) Fang <fanglu@meta.com>
```
  f48b6a03
- Update to Transformers `v4.56.2` (#24638) · 2a69ab48
  Harry Mellor authored Oct 01, 2025
```
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>
Co-authored-by: Isotr0py <mozf@mail2.sysu.edu.cn>
```
  2a69ab48
- [BugFix] Fix default kv-cache-dtype default for DeepseekV3.2 (#25988) · 8d7da92f
  Lucas Wilkinson authored Oct 01, 2025
```
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
```
  8d7da92f
- [Bugfix] Fix `__syncwarp` on ROCM (#25996) · e952eee6
  Zhewen Li authored Sep 30, 2025
  
  e952eee6
- [MM] Add text-only mode for Qwen3-VL (#26000) · 66bca9b8
  Roger Wang authored Sep 30, 2025
  
  66bca9b8
- Fix INT8 quantization error on Blackwell GPUs (SM100+) (#25935) · 99028fda
  Param authored Sep 30, 2025
```
Signed-off-by: padg9912 <phone.and.desktop@gmail.com>
```
  99028fda
- [Log] Optimize Log for FP8MOE (#25709) · 12449488
  Wentao Ye authored Sep 30, 2025
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  12449488
- Update launch_bounds_utils.h for correct compile on Multiple Cuda Arch - PTXAS... · a73f6491
  Salvatore Cena authored Oct 01, 2025
```
Update launch_bounds_utils.h for correct compile on Multiple Cuda Arch - PTXAS out of range Warning (#25843)
Signed-off-by: Salvatore Cena <cena@cenas.it>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
```
  a73f6491
- [Model] MTP fallback to eager for DeepSeek v32 (#25982) · 001e50c9
  Lucia Fang authored Sep 30, 2025
```
Signed-off-by: Lu Fang <fanglu@fb.com>
```
  001e50c9
30 Sep, 2025 23 commits
- [Misc] Make EP kernels install script support uv (#25785) · 96ebcaa3
  Lucas Wilkinson authored Sep 30, 2025
```
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
```
  96ebcaa3
- [gpt-oss] use vLLM instead of openai types for streaming (#25186) · 5db1870b
  Andrew Xia authored Sep 30, 2025
```
Signed-off-by: Andrew Xia <axia@meta.com>
Signed-off-by: Andrew Xia <axia@fb.com>
Co-authored-by: Andrew Xia <axia@fb.com>
```
  5db1870b
- [Docs] Remove API Reference from search index (#25949) · 2ce26b9b
  Harry Mellor authored Sep 30, 2025
```
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
```
  2ce26b9b
- Add explicit pooling classes for the Transformers backend (#25322) · a388252a
  Harry Mellor authored Sep 30, 2025
```
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
Co-authored-by: Isotr0py <mozf@mail2.sysu.edu.cn>
```
  a388252a
- [V1] [P/D] Add Support for KV Load Failure Recovery (#19330) · 9a9f48df
  David Ben-David authored Oct 01, 2025
```
Signed-off-by: David Ben-David <davidb@pliops.com>
Co-authored-by: David Ben-David <davidb@pliops.com>
```
  9a9f48df
- [Bench] Add DeepSeekV32 to MoE benchmark (#25962) · 67f3fb08
  Jee Jee Li authored Oct 01, 2025
```
Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
```
  67f3fb08
- [Llama4] [multimodal] Fix misplaced dtype cast of `cos_sin_cache` in... · 43b752c3
  cjackal authored Oct 01, 2025
```
[Llama4] [multimodal] Fix misplaced dtype cast of `cos_sin_cache` in `Llama4VisionRotaryEmbedding` (#25889)
Signed-off-by: cjackal <44624812+cjackal@users.noreply.github.com>
```
  43b752c3
- OffloadingConnector: Fix GPU block tracking bug (#25856) · cfd302db
  Or Ozeri authored Sep 30, 2025
```
Signed-off-by: Or Ozeri <oro@il.ibm.com>
```
  cfd302db
- [Docs] Add moe kernel features doc (#25297) · fb610ae6
  bnellnm authored Sep 30, 2025
```
Signed-off-by: Bill Nell <bnell@redhat.com>
Signed-off-by: bnellnm <49004751+bnellnm@users.noreply.github.com>
Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
```
  fb610ae6
- [Doc] Improve MM Pooling model documentation (#25966) · 2f652e6c
  Cyrus Leung authored Oct 01, 2025
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  2f652e6c
- [Bug] Fix AttributeError: 'QKVParallelLinear' object has no attribute 'orig_dtype' (#25958) · e6a226ef
  Wentao Ye authored Sep 30, 2025
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  e6a226ef
- [bugfix][deepseek] fix flashmla kernel selection (#25956) · a2e6fa7e
  youkaichao authored Oct 01, 2025
```
Signed-off-by: youkaichao <youkaichao@gmail.com>
```
  a2e6fa7e
- [Bugfix] Token type and position embeddings fail to be applied to `inputs_embeds` (#25922) · 9f1c4eca
  Cyrus Leung authored Oct 01, 2025
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  9f1c4eca
- [Bugfix] Fix accuracy issue of TRTLLM FP8 MOE and improve logging (#25895) · ef283548
  Pavani Majety authored Sep 30, 2025
```
Signed-off-by: Pavani Majety <pmajety@nvidia.com>
```
  ef283548
- [Bugfix][Model] Fix inference for Hunyuan dense models (#25354) · f4db5e6d
  Anion authored Sep 30, 2025
```
Signed-off-by: anion <1005128408@qq.com>
Signed-off-by: Anion <123177548+Anionex@users.noreply.github.com>
```
  f4db5e6d
- Add Hugging Face Inference Endpoints guide to Deployment docs (#25886) · 099aaee5
  Sergio Paniego Blanco authored Sep 30, 2025
```
Signed-off-by: sergiopaniego <sergiopaniegoblanco@gmail.com>
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
```
  099aaee5
- [Kernel][Moe Configs] Add more tuned triton configs for ExpertsInt8 and FP8 (#25858) · 35fe398c
  Asaf Joseph Gardin authored Sep 30, 2025
```
Signed-off-by: asafg <39553475+Josephasafg@users.noreply.github.com>
```
  35fe398c
- [Fix] Improve CPU backend compatibility for RISC-V (#25816) · bb6d4304
  ihb2032 authored Sep 30, 2025
```
Signed-off-by: lyd1992 <liuyudong@iscas.ac.cn>
Signed-off-by: ihb2032 <1355790728@qq.com>
```
  bb6d4304
- [CI] Move applicable tests to CPU (#24080) · bc546f76
  Reza Barazesh authored Sep 30, 2025
```
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
```
  bc546f76
- [NIXL] Add support for MLA caches with different latent dim (#25902) · 80608ba5
  Nicolò Lucchesi authored Sep 30, 2025
```
Signed-off-by: NickLucche <nlucches@redhat.com>
Signed-off-by: Chen Zhang <zhangch99@outlook.com>
Co-authored-by: Chen Zhang <zhangch99@outlook.com>
```
  80608ba5
- [perf] Use CPU tensor to reduce GPU->CPU sync (#25884) · e184c9c5
  Lehua Ding authored Sep 30, 2025
```
Signed-off-by: Lehua Ding <lehuading@tencent.com>
```
  e184c9c5
- [Model] Move `vision_feature_select_strategy` into `resolve_visual_encoder_outputs` (#25938) · d7e34b42
  Cyrus Leung authored Sep 30, 2025
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  d7e34b42
- [Bugfix][Model]fix ernie45 moe gate&bias dtype to float32 (#25936) · ef6e0e71
  CSWYF3634076 authored Sep 30, 2025
```
Signed-off-by: wangyafeng <wangyafeng@baidu.com>
```
  ef6e0e71