Commits · e23cacda35132f14d05ee236e8e5ebd8703e6940 · OpenDAS / vllm_cscc

29 Sep, 2025 1 commit
- [XPU]Fix xpu spec decoding UTs, avoid using cuda graph (#25847) · 143844fa
  Kunshang Ji authored Sep 29, 2025
```
Signed-off-by: Kunshang Ji <kunshang.ji@intel.com>
```
  143844fa
27 Sep, 2025 3 commits

[Bugfix] Merge MM embeddings by index instead of token IDs (#16229) · 27d7638b

Cyrus Leung authored Sep 27, 2025


Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Signed-off-by: NickLucche <nlucches@redhat.com>
Signed-off-by: Roger Wang <hey@rogerw.io>
Co-authored-by: NickLucche <nlucches@redhat.com>
Co-authored-by: Roger Wang <hey@rogerw.io>

27d7638b

[Multimodal][Speculative Decoding]Eagle Eagle3 mm support, enablement on qwen2.5vl (#22872) · f1d53d15

WeiQing Chen authored Sep 27, 2025


Signed-off-by: Junhong <liujunhong11@huawei.com>
Signed-off-by: Junhong Liu <98734602+LJH-LBJ@users.noreply.github.com>
Co-authored-by: Junhong <liujunhong11@huawei.com>
Co-authored-by: LJH-LBJ <98734602+LJH-LBJ@users.noreply.github.com>

f1d53d15

[Spec decode] automatically disable mm for text-only draft models (#25667) · 6f5c0931
Jonas M. Kübler authored Sep 27, 2025
```
Signed-off-by: Jonas Kuebler <kuebj@amazon.com>
```
6f5c0931

26 Sep, 2025 2 commits
- [spec decode] Consolidate speculative decode method name for MTP (#25232) · c70ac4b8
  qizixi authored Sep 26, 2025
```
Signed-off-by: zixi-qi <qizixi@meta.com>
```
  c70ac4b8
- [CI/Build] fix doc build warning: Failed to get 'name: description' pair (#25733) · 3edf87d2
  yitingdc authored Sep 26, 2025
```
Signed-off-by: yiting.jiang <yiting.jiang@daocloud.io>
```
  3edf87d2
25 Sep, 2025 2 commits

[Spec Decode] Add Batch Parallel Ngram. Upto 8x lower overhead. (#24986) · e71b8e21

Ekagra Ranjan authored Sep 25, 2025


Signed-off-by: Ekagra Ranjan <3116519+ekagra-ranjan@users.noreply.github.com>
Co-authored-by: Nick Hill <nhill@redhat.com>

e71b8e21

[Model] Add LongCat-Flash (#23991) · 845adb3e

XuruiYang authored Sep 25, 2025


Signed-off-by: yangxurui <yangxurui@meituan.com>
Co-authored-by: yangxurui <yangxurui@meituan.com>

845adb3e

24 Sep, 2025 2 commits

[V0 Deprecation] Remove unused classes in attention (#25541) · e6750d0b

Woosuk Kwon authored Sep 24, 2025


Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
Signed-off-by: Woosuk Kwon <woosuk@thinkingmachines.ai>

e6750d0b

[Spec Decode] Enable FlashInfer Spec Decoding (#25196) · c30b405b

Benjamin Chislett authored Sep 23, 2025


Signed-off-by: Benjamin Chislett <benjamin.chislett@centml.ai>
Signed-off-by: Benjamin Chislett <bchislett@nvidia.com>
Co-authored-by: lhsjohn <huashuoli@tencent.com>

c30b405b

23 Sep, 2025 2 commits

[Speculators][Speculative Decoding] Fix gpt-oss eagle3 accuracy issue (#25406) · d5944d51
jiahanc authored Sep 23, 2025
```
Signed-off-by: jiahanc <173873397+jiahanc@users.noreply.github.com>
```
d5944d51

[Core/DBO][2/N] Dual-Batch Overlap add DeepEP High Throughput support and Prefill support (#24845) · cc1dc7ed

Lucas Wilkinson authored Sep 23, 2025


Signed-off-by: Sage Moore <sage@neuralmagic.com>
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
Signed-off-by: yewentao256 <zhyanwentao@126.com>
Signed-off-by: Lucas Wilkinson <LucasWilkinson@users.noreply.github.com>
Signed-off-by: Tyler Michael Smith <tyler@neuralmagic.com>
Co-authored-by: Sage Moore <sage@neuralmagic.com>
Co-authored-by: yewentao256 <zhyanwentao@126.com>
Co-authored-by: Tyler Michael Smith <tyler@neuralmagic.com>

cc1dc7ed

22 Sep, 2025 1 commit
- Enable Eagle3 speculative decoding for GPT-OSS model (#25246) · 21467f9a
  Eldar Kurtić authored Sep 22, 2025
```
Signed-off-by: Eldar Kurtic <8884008+eldarkurtic@users.noreply.github.com>
```
  21467f9a
21 Sep, 2025 1 commit
- [V0 Deprecation] Remove V0 sampling metadata (#25345) · 1c3ffdbe
  Woosuk Kwon authored Sep 21, 2025
```
Signed-off-by: Woosuk Kwon <woosuk@thinkingmachines.ai>
```
  1c3ffdbe
18 Sep, 2025 1 commit
- [Spec Decode] Efficient padded speculation (#24539) · b7433ca1
  Benjamin Chislett authored Sep 18, 2025
```
Signed-off-by: Benjamin Chislett <bchislett@nvidia.com>
```
  b7433ca1
16 Sep, 2025 1 commit

[Core/DBO][1/N] Add Dual-Batch Overlap mechanism to VLLM (#23693) · 56793995

Sage Moore authored Sep 16, 2025


Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
Signed-off-by: Sage Moore <sage@neuralmagic.com>
Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
Signed-off-by: yewentao256 <zhyanwentao@126.com>
Co-authored-by: Lucas Wilkinson <lwilkins@redhat.com>
Co-authored-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
Co-authored-by: yewentao256 <zhyanwentao@126.com>
Co-authored-by: Lucas Wilkinson <LucasWilkinson@users.noreply.github.com>
Co-authored-by: Robert Shaw <114415538+robertgshaw2-redhat@users.noreply.github.com>

56793995

14 Sep, 2025 1 commit

[Spec Decoding]Support Spec Decoding Metrics in DP Mode (#24049) · 90f3f7d7

wuhang authored Sep 15, 2025


Signed-off-by: wuhang <wuhang6@huawei.com>
Co-authored-by: Robert Shaw <114415538+robertgshaw2-redhat@users.noreply.github.com>

90f3f7d7

11 Sep, 2025 2 commits
- Update Spec Decode metrics to include drafted and accepted token throughput (#24127) · 79ac59f3
  Andrew Xia authored Sep 11, 2025
```
Signed-off-by: Andrew Xia <axia@meta.com>
```
  79ac59f3
- Add the support for the qwen3 next model (a hybrid attention model). (#24526) · e93f4cc9
  Tao He authored Sep 11, 2025
```
Signed-off-by: Tao He <linzhu.ht@alibaba-inc.com>
Co-authored-by: Jee Jee Li <pandaleefree@gmail.com>
```
  e93f4cc9
01 Sep, 2025 2 commits
- [bugfix]fix MTP hidden states (#24056) · a344a5aa
  Lucia Fang authored Sep 01, 2025
```
Signed-off-by: Lu Fang <fanglu@fb.com>
```
  a344a5aa
- [Doc]: fix typos in Python comments (#24026) · 10728495
  Didier Durand authored Sep 01, 2025
```
Signed-off-by: Didier Durand <durand.didier@gmail.com>
```
  10728495
20 Aug, 2025 3 commits
- [FIXBUG ] Allow disabling rocm_aiter_fa backend for ROCm GPUs not compatible with AITER (#22795) · 3b11b26b
  JartX authored Aug 20, 2025
```
Signed-off-by: JartX <sagformas@epdcenter.es>
Signed-off-by: tjtanaa <tunjian.tan@embeddedllm.com>
Co-authored-by: tjtanaa <tunjian.tan@embeddedllm.com>
```
  3b11b26b
- [Misc] Add max_seq_len to CommonAttentionMetadata (#23216) · d6d13bd4
  Woosuk Kwon authored Aug 20, 2025
```
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
```
  d6d13bd4
- [Model][V1] Support Ernie MTP (#22169) · 7cd17e22
  xyxinyang authored Aug 20, 2025
```
Signed-off-by: zhouchong <zhouchong03@baidu.com>
Co-authored-by: zhouchong <zhouchong03@baidu.com>
```
  7cd17e22
19 Aug, 2025 1 commit
- [Spec Decode] Make `propose_draft_token_ids` non-blocking for lower TTFT (#23041) · c9b38be8
  Woosuk Kwon authored Aug 18, 2025
```
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
```
  c9b38be8
13 Aug, 2025 2 commits
- [Core] [N-gram SD Optimization][1/n] Propose tokens with a single KMP (#22437) · 31a500c8
  Jialin Ouyang authored Aug 13, 2025
```
Signed-off-by: Jialin Ouyang <Jialin.Ouyang@gmail.com>
```
  31a500c8
- [V1] Add tree drafting tests for eagle spec decoding (#22705) · d94e3026
  Giancarlo Delfin authored Aug 13, 2025
```
Signed-off-by: Giancarlo Delfin <gdelfin@meta.com>
```
  d94e3026
08 Aug, 2025 1 commit
- [ROCm] [V1] [SpecDec] Enable Speculative Decoding on ROCm V1 Engine (#21496) · 1ee5ead5
  TJian authored Aug 07, 2025
```
Signed-off-by: tjtanaa <tunjian.tan@embeddedllm.com>
```
  1ee5ead5
07 Aug, 2025 1 commit

[Attention] Support multiple attention metadata builders per kv_cache_spec +... · 1dc8a70b

Lucas Wilkinson authored Aug 06, 2025


[Attention] Support multiple attention metadata builders per kv_cache_spec  + proper local attention no hybrid kv cache fix (#21588)
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>

1dc8a70b

04 Aug, 2025 1 commit
- Add tree attention backend for v1 (part 1) (#20401) · aa7012eb
  Giancarlo Delfin authored Aug 03, 2025
```
Signed-off-by: Giancarlo Delfin <gdelfin@meta.com>
```
  aa7012eb
31 Jul, 2025 1 commit
- [Meta] Official Eagle mm support, first enablement on llama4 (#20788) · 9e0726e5
  zhiweiz authored Jul 31, 2025
```
Signed-off-by: morgendave <morgendave@gmail.com>
Co-authored-by: Roger Wang <hey@rogerw.me>
```
  9e0726e5
26 Jul, 2025 1 commit

Support encoder-only models without KV-Cache (#21270) · 1cd6eaba

Maximilien de Bayser authored Jul 26, 2025


Signed-off-by: Max de Bayser <maxdebayser@gmail.com>
Signed-off-by: Max de Bayser <mbayser@br.ibm.com>
Co-authored-by: Russell Bryant <rbryant@redhat.com>

1cd6eaba

17 Jul, 2025 1 commit
- [Attention] Refactor attention metadata builder interface (#20466) · 76b49444
  Lucas Wilkinson authored Jul 17, 2025
```
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
```
  76b49444
06 Jul, 2025 1 commit
- [V1] Support any head size for FlexAttention backend (#20467) · 9fb52e52
  Cyrus Leung authored Jul 07, 2025
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  9fb52e52
02 Jul, 2025 1 commit

[V1] LogitsProcessor programming model (#16728) · 48fb076c

afeldman-nm authored Jul 02, 2025


Signed-off-by: Nick Hill <nhill@redhat.com>
Signed-off-by: Andrew Feldman <afeldman@neuralmagic.com>
Signed-off-by: Andrew Feldman <afeldman@redhat.com>
Co-authored-by: Nick Hill <nhill@redhat.com>

48fb076c

25 Jun, 2025 1 commit
- [V1][Speculative Decoding] Fix DeepSeek MTP (#20022) · 8359f4c8
  cjackal authored Jun 26, 2025
```
Signed-off-by: cjackal <44624812+cjackal@users.noreply.github.com>
```
  8359f4c8
13 Jun, 2025 3 commits
- [CUDA] Enable full cudagraph for FlashMLA (#18581) · 3597b06a
  Luka Govedič authored Jun 13, 2025
```
Signed-off-by: luka <luka@neuralmagic.com>
```
  3597b06a
- [torch.compile] reorganize the cache directory to support compiling multiple models (#19064) · d70bc7c0
  youkaichao authored Jun 13, 2025
```
Signed-off-by: youkaichao <youkaichao@gmail.com>
```
  d70bc7c0
- [Bugfix] Fix EAGLE vocab embedding for multimodal target model (#19570) · c68698b3
  qizixi authored Jun 12, 2025
```
Signed-off-by: qizixi <qizixi@meta.com>
```
  c68698b3
06 Jun, 2025 1 commit
- [Bugfix] Fix EAGLE vocab embedding construction for Llama 70B (#19033) · 3465b87e
  Benjamin Chislett authored Jun 05, 2025
```
Signed-off-by: Benjamin Chislett <benjamin.chislett@centml.ai>
```
  3465b87e