Commits · 82531edbfb6b33e1c8667dea15c8622f011dcef0 · OpenDAS / vllm_cscc

"tests/v1/entrypoints/openai/test_completion.py" did not exist on "dd793d1de59b5efad25f4794b68cb935824c7a11"

16 Apr, 2026 1 commit
- [Model Runner V2][BugFix] fix num_sampled dtype for probabilistic rej… (#39951) · 343f6523
  Giancarlo Delfin authored Apr 15, 2026
```
Signed-off-by: Giancarlo Delfin <gdelfin@inferact.ai>
```
  343f6523
15 Apr, 2026 5 commits

[XPU] properly handle q_descale on XPU as quant query input not supported (#39676) · 3beb57a2
Yan Ma authored Apr 15, 2026
```
Signed-off-by: Yan Ma <yan.ma@intel.com>
Co-authored-by: Kunshang Ji <kunshang.ji@intel.com>
```
3beb57a2

[Metrics] Add request_id to FinishedRequestStats to enable correlation between... · 68be0f85

Csrayz authored Apr 15, 2026


[Metrics] Add request_id to FinishedRequestStats to enable correlation between metrics and requests (#39710)

Enables external `StatLogger` plugins to correlate per-request metrics
with request-level context. Also, this is a pre-requisite for Prometheus
exemplars in #30972.
Signed-off-by: Csrayz <33659823+Csrayz@users.noreply.github.com>

68be0f85

[kv_offload+HMA][3/N]: Remove block_size from KVEvents (#36644) · 235e1f93
Or Ozeri authored Apr 15, 2026
```
Signed-off-by: Or Ozeri <oro@il.ibm.com>
```
235e1f93

[Attention Backend] TurboQuant: 2-bit KV cache compression with 4x capacity (#38479) · f4b42df0

Vibhav Agarwal authored Apr 15, 2026


Signed-off-by: vibhavagarwal5 <vibhavagarwal5@gmail.com>
Signed-off-by: Michael Goin <mgoin64@gmail.com>
Co-authored-by: Xinyu Chen <xinyu1.chen@intel.com>
Co-authored-by: Michael Goin <mgoin64@gmail.com>

f4b42df0

[Model Runner V2] Disable piecewise cudagraph mode fallback for eagle draft decodes (#39773) · 3bfe55a0
Giancarlo Delfin authored Apr 14, 2026
```
Signed-off-by: Giancarlo Delfin <gdelfin@inferact.ai>
```
3bfe55a0

14 Apr, 2026 8 commits
- [Hybrid] Simplify accepted token counting in spec decode for hybrid models (#38372) · 507df79a
  Francesco Fusco authored Apr 15, 2026
  
  507df79a
- [Mamba] Flashinfer selective_state_update (#36162) · ecf5ff7c
  roikoren755 authored Apr 14, 2026
```
Signed-off-by: Roi Koren <roik@nvidia.com>
```
  ecf5ff7c
- [KV Offload] Unified memory layout for offloading workers (#37206) · be0c855e
  omerpaz95 authored Apr 14, 2026
```
Signed-off-by: omerpaz95 <omerpaz95@gmail.com>
Co-authored-by: Or Ozeri <oro@il.ibm.com>
```
  be0c855e
- [ROCm] Align AiterFlashAttentionImpl attn_type check with backend (#39119) · e64b39ea
  Andrew Barnes authored Apr 14, 2026
```
Signed-off-by: Bortlesboat <bortstheboat@gmail.com>
```
  e64b39ea
- [Core][Metrics] Remove unused `SchedulerStats.encoder_cache_usage` (#39693) · 906a8c15
  Mark McLoughlin authored Apr 14, 2026
```
Signed-off-by: Mark McLoughlin <markmc@redhat.com>
```
  906a8c15
- Measure encoder compile time seperate from llm backbone (#39240) · c08f3b2a
  Lucas Kabela authored Apr 14, 2026
```
Signed-off-by: Lucas Kabela <lucaskabela@meta.com>
```
  c08f3b2a
- [MM][Perf][CG] Support ViT full CUDA graph for Qwen3-VL video inference (#38061) · 80118853
  Shanshan Shen authored Apr 14, 2026
```
Signed-off-by: shen-shanshan <467638484@qq.com>
Signed-off-by: Shanshan Shen <87969357+shen-shanshan@users.noreply.github.com>
Co-authored-by: Roger Wang <hey@rogerw.io>
```
  80118853
- [Core][Metrics][BugFix] Replace num_cached_tokens/num_external_computed_tokens... · d3af8c18
  Mark McLoughlin authored Apr 14, 2026
```
[Core][Metrics][BugFix] Replace num_cached_tokens/num_external_computed_tokens with PrefillStats (#37460)

Related to `Counters can only be incremented by non-negative amounts`
error with the `vllm:prompt_tokens_by_source_total` metric.
Signed-off-by: Mark McLoughlin <markmc@redhat.com>
Co-authored-by: Or Ozeri <or@ozery.com>
```
  d3af8c18
13 Apr, 2026 3 commits

[Model Runner V2] Add full cuda graph support for eagle prefill (#37588) · ccf90ba7
Giancarlo Delfin authored Apr 13, 2026
```
Signed-off-by: Giancarlo Delfin <gdelfin@inferact.ai>
```
ccf90ba7

[Core][Metrics] expose waiting request breakdown via labeled metric (capacity/deferred) (#38435) · 5c18b961

mukesh-hai authored Apr 13, 2026


Signed-off-by: Mukesh Baphna <mukesh@hippocraticai.com>
Signed-off-by: Mark McLoughlin <markmc@redhat.com>
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-authored-by: Mark McLoughlin <markmc@redhat.com>

5c18b961

[Bug] Fix rocm sparse attn indexer issue (#39225) · ccd0d1d9

Wentao Ye authored Apr 13, 2026


Signed-off-by: yewentao256 <zhyanwentao@126.com>
Co-authored-by: TJian <tunjian.tan@embeddedllm.com>

ccd0d1d9

12 Apr, 2026 3 commits
- [Core][Metrics] Remove `vllm:prompt_tokens_recomputed` metric (#38709) · 72ff142c
  Mark McLoughlin authored Apr 12, 2026
```
Signed-off-by: Mark McLoughlin <markmc@redhat.com>
```
  72ff142c
- [HMA] [KVEvent] Enable GPU-side KV events for HMA (#37688) · cc07dad7
  Martin Hickey authored Apr 12, 2026
```
Signed-off-by: Martin Hickey <martin.hickey@ie.ibm.com>
Co-authored-by: Or Ozeri <or@ozery.com>
```
  cc07dad7
- Support FP8 KVCache on XPU (#37731) · 639402f5
  Xinyu Chen authored Apr 12, 2026
```
Signed-off-by: Xinyu Chen <xinyu1.chen@intel.com>
Co-authored-by: Kunshang Ji <kunshang.ji@intel.com>
```
  639402f5
11 Apr, 2026 1 commit

Fix the order of _free_encoder_inputs (#38907) · 8d0aabdd

Tianyu Guo authored Apr 11, 2026


Signed-off-by: Tianyu Guo <guoty9@mail2.sysu.edu.cn>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>

8d0aabdd

10 Apr, 2026 14 commits
- Add Gemma4 Eagle3 support (#39450) · e7cfd7c5
  Fynn Schmitt-Ulms authored Apr 10, 2026
```
Signed-off-by: Rahul-Tuli <rtuli@redhat.com>
Signed-off-by: Fynn Schmitt-Ulms <fschmitt@redhat.com>
Co-authored-by: Rahul-Tuli <rtuli@redhat.com>
Co-authored-by: Claude <noreply@anthropic.com>
```
  e7cfd7c5
- [Bugfix] Fix FlashInfer crash with kv_cache_dtype_skip_layers (#39002) · e816a881
  yzong-rh authored Apr 10, 2026
```
Signed-off-by: Yifan Zong <yzong@redhat.com>
```
  e816a881
- [Model Runner V2] support auto resolve cudagraph mode/sizes based on attn backend (#32936) · 8f121f78
  zhrrr authored Apr 10, 2026
```
Signed-off-by: zhuhaoran <zhuhaoran.zhr@alibaba-inc.com>
```
  8f121f78
- [Model] Implement LoRA support for Qwen3ASRForConditionalGeneration (#37247) · 8d0f908b
  Peter Nguyen authored Apr 10, 2026
```
Signed-off-by: Peter Nguyen <petern0408@gmail.com>
```
  8d0f908b
- [Performance] Remove unnecessary zero-fill of MLA decode output tensor in Aiter backend (#37539) · f976e3b9
  xaguilar-amd authored Apr 10, 2026
```
Signed-off-by: xaguilar-amd <xaguilar@amd.com>
```
  f976e3b9
- [Bugfix] Fix V1 dummy run writing NaN to KV cache null block (#39444) · ad720aef
  Elvir Crnčević authored Apr 10, 2026
```
Signed-off-by: Elvir Crncevic <elvircrn@gmail.com>
Co-authored-by: Claude Sonnet 4 <noreply@anthropic.com>
```
  ad720aef
- [compile] Allow strings in custom ops without regressing compilation times (#38123) · f44afef6
  Richard Zou authored Apr 10, 2026
```
Signed-off-by: Richard Zou <zou3519@gmail.com>
```
  f44afef6
- [Perf] Reduce H2D pageable memory copies (#38794) · 49d20346
  jackwang2120 authored Apr 10, 2026
```
Signed-off-by: jackcfwang <jackcfwang@tencent.com>
Co-authored-by: Isotr0py <mozf@mail2.sysu.edu.cn>
```
  49d20346
- [Core] Change max_model_len in EngineCoreReadyResponse to be non-None (#39442) · ef076c1b
  Nick Hill authored Apr 09, 2026
```
Signed-off-by: Nick Hill <nickhill123@gmail.com>
```
  ef076c1b
- [feat]: make DCP error msg clearer (#28443) · 3dd60971
  Srreyansh Sethi authored Apr 09, 2026
```
Signed-off-by: vnadathur <glvikramn@gmail.com>
Signed-off-by: WorldExplored <srreyansh.sethi@gmail.com>
Signed-off-by: Srreyansh Sethi <107075589+WorldExplored@users.noreply.github.com>
Co-authored-by: vnadathur <glvikramn@gmail.com>
Co-authored-by: vnadathur <236933696+vnadathur@users.noreply.github.com>
```
  3dd60971
- [KV Offload] Implement `shutdown()` in `OffloadingConnector` and related classes (#39182) · a5b17fba
  Ronen Schaffer authored Apr 10, 2026
```
Signed-off-by: Ronen Schaffer <ronen.schaffer@ibm.com>
```
  a5b17fba
- Add EXAONE-4.5 (#39388) · e7a1387e
  Kyungmin Lee authored Apr 10, 2026
```
Signed-off-by: lkm2835 <lkm2835@gmail.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
```
  e7a1387e
- feat(cpu): add CPU support for draft model speculative decoding (#32662) · 445a2a4d
  Ganesh R authored Apr 10, 2026
```
Signed-off-by: R <Ganesh.R@amd.com>
```
  445a2a4d
- [Feature] Add auto-detection for reasoning_config when only reasoning_parser is set (#38214) · ecbfbb8d
  Chauncey authored Apr 10, 2026
```
Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
```
  ecbfbb8d
09 Apr, 2026 5 commits
- [Spec Decode] fix returning size mismatch on extract hidden states proposer (#38610) · edee9651
  zzaebok authored Apr 10, 2026
```
Signed-off-by: Jaebok Lee <jaebok9541@naver.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
```
  edee9651
- [XPU] check is_xccl_available before oneccl warmup (#39302) · a8134aef
  Xinyu Chen authored Apr 10, 2026
```
Signed-off-by: Xinyu Chen <xinyu1.chen@intel.com>
```
  a8134aef
- [Model Runner V2] Fix flex attention kv blocks calculation issue (#39353) · 56e19d7e
  Wentao Ye authored Apr 09, 2026
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  56e19d7e
- [Performance Improvement] Update `batched_count_greater_than` to handle batch... · a8c6ee9b
  Lucas Kabela authored Apr 09, 2026
```
[Performance Improvement] Update `batched_count_greater_than` to handle batch size 1 without recompile (#38933)
Signed-off-by: Lucas Kabela <lucaskabela@meta.com>
Co-authored-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
```
  a8c6ee9b
- [UX] Improve error message for MM input too long (#39409) · 54d244f2
  Cyrus Leung authored Apr 09, 2026
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  54d244f2