Commits · ec5f9c6269c18e88cd3141be56369cbd23cf4d64 · zhaoyu6 / sglang

"tests/vscode:/vscode.git/clone" did not exist on "c9ad3605a78bb06e3729d7c82f4e6a700ccf584a"

06 Jul, 2025 2 commits
- chore: bump v0.4.9 (#7802) · ec5f9c62
  Yineng Zhang authored Jul 05, 2025
  
  ec5f9c62
- chore: upgrade sgl-kernel v0.2.4 (#7801) · 62f5522f
  Yineng Zhang authored Jul 05, 2025
  
  62f5522f
05 Jul, 2025 12 commits
- Add treemask mode to build_eagle_tree & release sgl-kernel 0.2.3 (#7756) · 5589b750
  Lianmin Zheng authored Jul 05, 2025
```
Co-authored-by: Pranjal Shankhdhar <pranjal.ssh@gmail.com>
```
  5589b750
- [fix] fix misusing of is_cuda (#7790) · c04a8a82
  JieXin Liang authored Jul 05, 2025
  
  c04a8a82
- Fix incorrect spec_num_draft_tokens in draft_extend (#7757) · 6c903611
  Cheng Wan authored Jul 05, 2025
  
  6c903611
- chore: upgrade sgl-kernel v0.2.3 (#7786) · 77cfea68
  Yineng Zhang authored Jul 05, 2025
  
  77cfea68
- DP Attention with Auto DeepEP Dispatch (#7222) · 8fc910db
  Cheng Wan authored Jul 05, 2025
  
  8fc910db
- Embedding parallel by attn_tp (#7623) · ef8a29c4
  Gang Chen authored Jul 05, 2025
  
  ef8a29c4
- add model: qwen2-audio (#7596) · 83646089
  Leng Yue authored Jul 04, 2025
  
  83646089
- saving hidden_states.clone() (#7705) · cb432f17
  Cheng Wan authored Jul 04, 2025
  
  cb432f17
- [feat] Support EAGLE3 for Qwen (#7745) · 1964c325
  Ximingwang-09 authored Jul 05, 2025
```
Co-authored-by: 纬杭 <ximing.wxm@antgroup.com>
Co-authored-by: zyksir <zyksir@outlook.com>
```
  1964c325
- [Fix] Alloc return type error (#7778) · af564774
  Caproni authored Jul 05, 2025
```
Signed-off-by: Capronir <839972205@qq.com>
```
  af564774
- [RL] add pause and continue generation for async rl training (#7419) · af46f299
  Zilin Zhu authored Jul 05, 2025
  
  af46f299
- [RL] Add --nccl-port to prevent port conflict (#7418) · 16a6b1d8
  Zilin Zhu authored Jul 05, 2025
  
  16a6b1d8
04 Jul, 2025 4 commits
- Move mem_fraction_static adjustment for multimodal models to `server_args.py`... · 14229ccf
  Lianmin Zheng authored Jul 04, 2025
```
Move mem_fraction_static adjustment for multimodal models to `server_args.py` & Fix session control & Other cleanups (#7748)
```
  14229ccf
- refactor llama4 dp attention logic (#7729) · 8c298031
  Yi Zhang authored Jul 04, 2025
  
  8c298031
- Add V2-lite model test (#7390) · 4de03953
  YanbingJiang authored Jul 04, 2025
```
Co-authored-by: DiweiSun <105627594+DiweiSun@users.noreply.github.com>
```
  4de03953
- Remove type conversion and fix id map in topk (#7759) · 8b1942c6
  Ke Bao authored Jul 04, 2025
  
  8b1942c6
03 Jul, 2025 20 commits
- fuse renormal into moe topk softmax kernel python code (#7751) · 489934be
  Yi Zhang authored Jul 04, 2025
```
Co-authored-by: ispobock <ispobaoke@gmail.com>
Co-authored-by: zhyncs <me@zhyncs.com>
```
  489934be
- [fix] put cpu in the first priority in get_device() (#7752) · 6840a7bb
  JieXin Liang authored Jul 04, 2025
  
  6840a7bb
- [Bug] add flashinfer bool check for fusedmoe in Qwen moe models (#7723) · c01a1df5
  yilian49 authored Jul 03, 2025
  
  c01a1df5
- feat: use D2D instead of H2H in pp (#7673) · 00991723
  TianyuZhang1214 authored Jul 04, 2025
```
Co-authored-by: alpha-baby <fujianhao1997@qq.com>
```
  00991723
- [optimize] add two stream norm for qwen3 (#7740) · 264dc6e7
  Yi Zhang authored Jul 04, 2025
```
Co-authored-by: ispobock <ispobaoke@gmail.com>
```
  264dc6e7
- support qwen3 dense model dp attention (#7681) · 646cef2e
  Yi Zhang authored Jul 04, 2025
  
  646cef2e
- [CPU] support the case where num_attention_heads or intermediate_size is not... · 1dce6c48
  Chunyuan WU authored Jul 04, 2025
```
[CPU] support the case where num_attention_heads or intermediate_size is not divisible by the TP size (#6771)
```
  1dce6c48
- [CPU] refine CPU integration code (#7647) · 9fcc9a80
  Chunyuan WU authored Jul 04, 2025
  
  9fcc9a80
- [fix] fix dsv3_router_gemm filter (#7750) · ac49dac0
  JieXin Liang authored Jul 04, 2025
  
  ac49dac0
- Ascend attention backend(PA&MLA) (#7722) · 1e0e5497
  ronnie_zheng authored Jul 03, 2025
```
Co-authored-by: Maksim <makcum888e@mail.ru>
Co-authored-by: VDV1985 <vladdv85@mail.ru>
```
  1e0e5497
- fix dsv3 fused proj check (#7738) · b5822651
  AniZpZ authored Jul 03, 2025
  
  b5822651
- [CI][Router] Fix bench_one_batch_server for pd router test (#7731) · 2ff572e2
  Shangming Cai authored Jul 03, 2025
```
Signed-off-by: Shangming Cai <caishangming@linux.alibaba.com>
```
  2ff572e2
- fix awq and dsv3 fused gemm compatible (#7735) · 84f2e4a0
  AniZpZ authored Jul 03, 2025
  
  84f2e4a0
- [CPU] fix all_reduce and all_gather (#6770) · 8f844db6
  Chunyuan WU authored Jul 03, 2025
```
Co-authored-by: blzheng <beilei.zheng@intel.com>
```
  8f844db6
- Fix num_tokens_pre_allocated in disaggregation log (#7714) · 1bebd315
  Ziming Huang authored Jul 03, 2025
  
  1bebd315
- Support updating weights at once by stopping all requests (#6698) · d3c275b1
  Albert authored Jul 03, 2025
```
Signed-off-by: Tianyu Zhou <albert.zty@antgroup.com>
Co-authored-by: Zilin Zhu <zhuzilinallen@gmail.com>
```
  d3c275b1
- [CPU] Bind threads and numa node for each TP rank (#6549) · 40e5cb7a
  Chunyuan WU authored Jul 03, 2025
```
Co-authored-by: srinarayan-srikanthan <srinarayan.srikanthan@intel.com>
```
  40e5cb7a
- [b200] support trt-llm allreduce fuse rms_norm_add kernel (#7621) · 8e64140e
  Xiaoyu Zhang authored Jul 03, 2025
  
  8e64140e
- [RL] support update_weights_from_distributed with different group and multiple weights (#7292) · 0626f678
  Zilin Zhu authored Jul 03, 2025
  
  0626f678
- [RL] add --skip-warmup (#7416) · 09e699bb
  Zilin Zhu authored Jul 03, 2025
  
  09e699bb
02 Jul, 2025 2 commits

Apply dsv3 router gemm kernel for deepseek-r1 fp4 (#7677) · 88f484ce
Baizhou Zhang authored Jul 02, 2025

88f484ce

[1/n] apply wna16marlin kernel in moe weight only quantization (#7683) · 8e03b641

AniZpZ authored Jul 02, 2025


Co-authored-by: 晟海 <huangtingwei.htw@antgroup.com>
Co-authored-by: yych0745 <1398089567@qq.com>
Co-authored-by: HandH1998 <1335248067@qq.com>
Co-authored-by: 弋云 <yiyun.wyt@antgroup.com>
Co-authored-by: walker-ai <2398833647@qq.com>

8e03b641