Commits · b9c2a565ccfd5cd865ebca47cfb7ef0f7a8ba76c · OpenDAS / vllm_cscc

22 Feb, 2026 4 commits
- [Spec Decode] Defer clearing KV connector metadata for EAGLE3 speculative... · b9c2a565
  qizixi authored Feb 22, 2026
```
[Spec Decode] Defer clearing KV connector metadata for EAGLE3 speculative decode + prefill / decode disagg setup (#34529)
Signed-off-by: qizixi <qizixi@meta.com>
Co-authored-by: Lu Fang <30275821+houseroad@users.noreply.github.com>
```
  b9c2a565
- [Model Runner V2] Enable CUDA graph for Eagle3 (#35040) · 2cbf9656
  Woosuk Kwon authored Feb 21, 2026
```
Signed-off-by: Woosuk Kwon <woosuk@inferact.ai>
```
  2cbf9656
- Fix apply_top_k_top_p_triton called by non-cuda logits Tensor (#35030) · 30132cd1
  Xiao Li authored Feb 21, 2026
```
Signed-off-by: Xiao Li <ilx@meta.com>
```
  30132cd1
- [Model Runner V2] Support attention group (#35036) · b71fbd06
  Woosuk Kwon authored Feb 21, 2026
```
Signed-off-by: Woosuk Kwon <woosuk@inferact.ai>
```
  b71fbd06
21 Feb, 2026 7 commits
- [Model Runner V2] Support Eagle3 (no CUDA graph) (#35029) · a4047d4e
  Woosuk Kwon authored Feb 21, 2026
```
Signed-off-by: Woosuk Kwon <woosuk@inferact.ai>
```
  a4047d4e
- [Core] Minor structured-output related scheduler optimization (#34765) · 820d7815
  Nick Hill authored Feb 21, 2026
```
Signed-off-by: Nick Hill <nickhill123@gmail.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
```
  820d7815
- [ROCM] Optimize ROCM_AITER_FA spec decode eagle performance (#34541) · 2aab2bb5
  jennyyyyzhen authored Feb 20, 2026
```
Signed-off-by: jennyyyyzhen <yzhen@hmc.edu>
```
  2aab2bb5
- [ROCm][AITER] Fix aiter paged_attention_v1 decode for sliding window and head_size < 64 (#34570) · cf93c1a1
  Andreas Karatzas authored Feb 20, 2026
```
Signed-off-by: Andreas Karatzas <akaratza@amd.com>
```
  cf93c1a1
- [feat] Add per-block extra_keys to KV events (#33304) · a0fe7ea2
  zhongdaor-nv authored Feb 20, 2026
```
Signed-off-by: zhongdaor-nv <zhongdaor@nvidia.com>
Co-authored-by: Nicolò Lucchesi <nlucches@redhat.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
```
  a0fe7ea2
- [Kernel] Optimize sample_recovered_tokens_kernel (#34974) · 7a5adad4
  Xin Yang authored Feb 20, 2026
```
Signed-off-by: Xin Yang <xyangx@amazon.com>
```
  7a5adad4
- Support prompt_embeds for pooling requests in output processor (#34904) · 59c62332
  Li authored Feb 20, 2026
```
Signed-off-by: Li Zhang <lzhanga@amazon.com>
Co-authored-by: Li Zhang <lzhanga@amazon.com>
```
  59c62332
20 Feb, 2026 7 commits
- [CI] Revert PRs 34818 and 33600 (#34979) · aaefc58e
  Lucas Wilkinson authored Feb 20, 2026
  
  aaefc58e
- [perf] Avoid dtype promotion sync in mamba_get_block_table_tensor (#34870) · 1fe46216
  Huamin Li authored Feb 20, 2026
```
Signed-off-by: Huamin Li <3ericli@gmail.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
```
  1fe46216
- [BUGFIX] Fix `_dummy_run` missing `prepare_inputs_event` synchronization (#34866) · 59965aff
  Vadim Gimpelson authored Feb 20, 2026
```
Signed-off-by: Vadim Gimpelson <vadim.gimpelson@gmail.com>
```
  59965aff
- [Bugfix][Hardware][AMD] Fix ROCM_AITER_FA speculative decoding support (#32877) · 8de7c636
  Kevin McKay authored Feb 20, 2026
```
Signed-off-by: c0de128 <kevin.mckay@outlook.com>
Co-authored-by: Lu Fang <30275821+houseroad@users.noreply.github.com>
```
  8de7c636
- [Refactor] Implement output type check in LLM (#34794) · ac900c89
  Cyrus Leung authored Feb 20, 2026
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  ac900c89
- [Core] Fix state names in pause_scheduler() (#34840) · 76df6072
  Mark McLoughlin authored Feb 20, 2026
```
Signed-off-by: Mark McLoughlin <markmc@redhat.com>
```
  76df6072
- [Model Runner V2] Minor CPU optimizations (#34856) · 40b2f1c3
  Nick Hill authored Feb 19, 2026
```
Signed-off-by: Nick Hill <nickhill123@gmail.com>
```
  40b2f1c3
19 Feb, 2026 2 commits

[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

[Model Runner V2] Use FP32 for Gumbel Noise (#34854) · 5fcb0cdd
Woosuk Kwon authored Feb 18, 2026
```
Signed-off-by: Woosuk Kwon <woosuk@inferact.ai>
```
5fcb0cdd

18 Feb, 2026 8 commits
- [Model Runner V2] Remove unnecessary copies in PW CUDA graph capture (#34849) · c878b43b
  Woosuk Kwon authored Feb 18, 2026
```
Signed-off-by: Woosuk Kwon <woosuk@inferact.ai>
```
  c878b43b
- [Model Runner V2] support piecewise & mixed cudagraph (#32771) · 11d3976b
  zhrrr authored Feb 19, 2026
```
Signed-off-by: zhuhaoran <zhuhaoran.zhr@alibaba-inc.com>
```
  11d3976b
- [Model Runner V2] Minor simplification for DCP (#34786) · 95be2a7f
  Woosuk Kwon authored Feb 18, 2026
```
Signed-off-by: Woosuk Kwon <woosuk@inferact.ai>
```
  95be2a7f
- [Model Runner V2] Avoid prepare prefill kernel launch overhead (#34780) · c50e105a
  Nick Hill authored Feb 18, 2026
```
Signed-off-by: Nick Hill <nickhill123@gmail.com>
```
  c50e105a
- [Renderer] Deprecate code paths for old input processing (#34775) · a766b303
  Cyrus Leung authored Feb 18, 2026
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  a766b303
- [Model Runner V2] A bit more PP simplification (#34766) · a49ea5a5
  Nick Hill authored Feb 17, 2026
```
Signed-off-by: Nick Hill <nickhill123@gmail.com>
```
  a49ea5a5
- [Attention] Refactor `check_and_update_config` (#33600) · 77431529
  Matthew Bonanni authored Feb 17, 2026
```
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
```
  77431529
- [Feature] Decode Context Parallel support for GPU model runner v2 (#34179) · ab33d2a6
  Wentao Ye authored Feb 17, 2026
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  ab33d2a6
17 Feb, 2026 12 commits
- [Model Runner V2] Further simplification for PP (#34724) · be3af2d2
  Woosuk Kwon authored Feb 17, 2026
```
Signed-off-by: Woosuk Kwon <woosuk@inferact.ai>
```
  be3af2d2
- [Kernel] Triton-based Top-k and Top-p sampler kernels (#33538) · c656ba3b
  Jongseok Park authored Feb 17, 2026
```
Signed-off-by: js_park <cakeng@naver.com>
Signed-off-by: Jongseok Park <37990712+cakeng@users.noreply.github.com>
Signed-off-by: Sunga Kim <sunga.kim@berkeley.edu>
Signed-off-by: Nick Hill <nickhill123@gmail.com>
Co-authored-by: Sunga Kim <sunga.kim@berkeley.edu>
Co-authored-by: Nick Hill <nickhill123@gmail.com>
```
  c656ba3b
- [Bugfix][MTP][Sparse MLA] Allow sparse MLA with MTP to run with FULL cudagraphs (#34457) · dc5fa77a
  Matthew Bonanni authored Feb 17, 2026
```
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
Co-authored-by: Jee Jee Li <pandaleefree@gmail.com>
```
  dc5fa77a
- [Renderer] Move InputPreprocessor into Renderer (2/2) (#34560) · 574fe752
  Cyrus Leung authored Feb 17, 2026
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  574fe752
- Fix docs build warning (#34686) · 28bffe94
  Harry Mellor authored Feb 17, 2026
```
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
```
  28bffe94
- [Ray] Propagate third-party env vars to Ray workers via prefix matching (#34383) · c464b573
  kourosh hakhamaneshi authored Feb 17, 2026
```
Signed-off-by: Kourosh Hakhamaneshi <kourosh@anyscale.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
```
  c464b573
- [Model Runner V2] Minor refactoring for penalties (#34662) · d00df624
  Woosuk Kwon authored Feb 16, 2026
```
Signed-off-by: Woosuk Kwon <woosuk@inferact.ai>
```
  d00df624
- [Model Runner V2] Minor simplification for BadWordsState (#34669) · 9752da9d
  Woosuk Kwon authored Feb 16, 2026
```
Signed-off-by: Woosuk Kwon <woosuk@inferact.ai>
```
  9752da9d
- [Model Runner V2] Minor cleanup for PP (#34666) · 04925b22
  Woosuk Kwon authored Feb 16, 2026
```
Signed-off-by: Woosuk Kwon <woosuk@inferact.ai>
```
  04925b22
- [Model Runner V2] Fix unintended CPU-GPU sync in make_dummy (#34667) · d74278fb
  Woosuk Kwon authored Feb 16, 2026
```
Signed-off-by: Woosuk Kwon <woosuk@inferact.ai>
```
  d74278fb
- [Core] Pipeline Parallel support for Model Runner V2 (#33960) · 9a8853f7
  zhanqiuhu authored Feb 16, 2026
```
Signed-off-by: Zhanqiu Hu <zh338@cornell.edu>
```
  9a8853f7
- [Model Runner V2] support bad_words sampling param (#33433) · 387a1898
  zhrrr authored Feb 17, 2026
```
Signed-off-by: zhuhaoran <zhuhaoran.zhr@alibaba-inc.com>
Signed-off-by: Woosuk Kwon <woosuk@inferact.ai>
Co-authored-by: Woosuk Kwon <woosuk@inferact.ai>
```
  387a1898