Commits · a2424068eca21d2b9c0fdfcfdccd588ab8dc87a3 · change / sglang · GitLab

11 Sep, 2025 16 commits
- add try catch for quant config hf download (#10340) · a2424068
  gongwei-130 authored Sep 11, 2025
  
  a2424068
- [LongCat] Optimize zero_experts_compute_triton by changing mask (#10303) · c5d2b01c
  zk-lover authored Sep 12, 2025
  
  c5d2b01c
- fix: trtllm-gen attention take zero-init workspace (#10330) · 70c0c1f9
  eigen authored Sep 11, 2025
  
  70c0c1f9
- add h20 qwen3 next config (#10264) · ab795ae8
  Yi Zhang authored Sep 12, 2025
```
Co-authored-by: cao1zhg <114661107+cao1zhg@users.noreply.github.com>
```
  ab795ae8
- [Qwen3-Next] switch to triton and cache conv states to accelerate MTP from 300... · 6c18ab46
  Stefan He authored Sep 11, 2025
```
[Qwen3-Next] switch to triton and cache conv states to accelerate MTP from 300 tok/s to 341 tok/s (#10335)
Co-authored-by: Binyao Jiang <byjiang1996@gmail.com>
```
  6c18ab46
- [bugfix] fix norm type error in qwen3_next model (#10322) · 4a0e0be2
  cao1zhg authored Sep 12, 2025
```
Co-authored-by: caoyizhong.cyz <caoyizhong.cyz@alibaba-inc.com>
Co-authored-by: Yi Zhang <1109276519@qq.com>
```
  4a0e0be2
- [Minor] Improve the style of server args (#10328) · 64f296f8
  Lianmin Zheng authored Sep 11, 2025
  
  64f296f8
- [Auto Sync] Update parallel_state.py (20250911) (#10326) · 956d805d
  Lianmin Zheng authored Sep 11, 2025
```
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Byron Hsu <byronhsu1230@gmail.com>
```
  956d805d
- Qwen3-Next support (#10233) · 30c6e1f5
  Yi Zhang authored Sep 11, 2025
```
Co-authored-by: cao1zhg <114661107+cao1zhg@users.noreply.github.com>
Co-authored-by: ispobock <ispobaoke@gmail.com>
Co-authored-by: Binyao Jiang <byjiang1996@gmail.com>
Co-authored-by: hebiao064 <hebiaobuaa@gmail.com>
Co-authored-by: Lifu Huang <lifu.hlf@gmail.com>
Co-authored-by: qingquansong <ustcsqq@gmail.com>
Co-authored-by: Yaoyao Ding <dingyaoyao.cs@gmail.com>
Co-authored-by: Ke Bao <ISPObaoke@163.com>
Co-authored-by: Minglei Zhu <mingleizhu1122@gmail.com>
```
  30c6e1f5
- chore: upgrade v0.3.9.post2 sgl-kernel (#10297) · bfe01a5e
  Yineng Zhang authored Sep 11, 2025
  
  bfe01a5e
- Revert "Fix flashinfer version in sgl-kernel (#10135)" (#10310) · de15d140
  Yineng Zhang authored Sep 11, 2025
  
  de15d140
- [fix CI] Fix logical condition in fused MoE layer for compressed tensor quantization (#10299) · 37367da6
  Xiaoyu Zhang authored Sep 11, 2025
  
  37367da6
- [CPU] fix OOM when mem-fraction is not set (#9090) · ef959d7b
  Zaili Wang authored Sep 11, 2025
  
  ef959d7b
- add flash linear attention triton kernel (#10239) · dc491b39
  Yi Zhang authored Sep 11, 2025
  
  dc491b39
- [Feature] Support DeepEP normal & Redundant Experts on NPU (#9881) · 5b64f006
  Even Zhou authored Sep 11, 2025
  
  5b64f006
- Revert "[1/2] Optimizations and refactors about quant kernel (#9534)" (#10292) · 6d55f60e
  Yineng Zhang authored Sep 10, 2025
  
  6d55f60e
10 Sep, 2025 15 commits
- [Auto Sync] Update serving_base.py, serving_chat.py, servin... (20250910) (#10282) · 033b75f5
  Lianmin Zheng authored Sep 10, 2025
```
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: cctry <shiyang@x.ai>
```
  033b75f5
- Feat: support disable tool parser (#10184) · f3b5db6e
  Xinyuan Tong authored Sep 10, 2025
  
  f3b5db6e
- pass a_scale from fp8 quant result instead of hard code to 1.0f (#10241) · 2286e85e
  Rain Jiang authored Sep 10, 2025
```
Co-authored-by: Yichen Wang <yichen.wang@bytedance.com>
Co-authored-by: Jinwu Guo <641876696@qq.com>
```
  2286e85e
- Add tests to AMD CI for MI35x (#9662) · 91b3555d
  Hubert Lu authored Sep 10, 2025
```
Co-authored-by: Sai Enduri <saimanas.enduri@amd.com>
```
  91b3555d
- add dual stream for qwen2_moe (#10252) · 9e2f7252
  Yi Zhang authored Sep 11, 2025
  
  9e2f7252
- [Bugfix] Fix Weightloading for the original nvidia/Deepseek-R1-FP4 checkpoint (#9940) · 21176b00
  Pavani Majety authored Sep 10, 2025
```
Signed-off-by: Pavani Majety <pmajety@nvidia.com>
Co-authored-by: Yineng Zhang <me@zhyncs.com>
Co-authored-by: fzyzcjy <5236035+fzyzcjy@users.noreply.github.com>
```
  21176b00
- [1/2] Refactor LoRA to support backend-specific batch preprocessing. (#10251) · 94100294
  Lifu Huang authored Sep 10, 2025
  
  94100294
- [Auto Sync] Update io_struct.py (20250910) (#10262) · 27760fc1
  Lianmin Zheng authored Sep 10, 2025
```
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Kan Wu <wukanustc@gmail.com>
```
  27760fc1
- Fix assertion typo in tp_worker.py (#9954) · 0ac809de
  Seunggeun Cho authored Sep 10, 2025
  
  0ac809de
- Fix: the default choice is wrong for flashinfer mxfp4 moe precision (#10253) · 737d73ed
  Yiyu Liu authored Sep 10, 2025
  
  737d73ed
- [UT for RL] Add UT to cover release/resume memory case for moe model (#8803) · dccf52f9
  ryang authored Sep 10, 2025
  
  dccf52f9
- make --speculative-draft-model an alias of --speculative-draft-model-path (#10246) · 676a7b51
  Lianmin Zheng authored Sep 09, 2025
  
  676a7b51
- refactor(InternVL): Use gpu to preprocess the input image (#9795) · 15f99347
  Kevin Tuan authored Sep 10, 2025
  
  15f99347
- Revert "chore: upgrade v0.3.9 sgl-kernel" (#10245) · bcf1955f
  Lianmin Zheng authored Sep 09, 2025
  
  bcf1955f
- [Auto Sync] Update collector.py, startup_func_log_and_timer... (20250910) (#10242) · a06bf664
  Lianmin Zheng authored Sep 09, 2025
```
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: cctry <shiyang@x.ai>
```
  a06bf664
09 Sep, 2025 9 commits
- [Auto Sync] Update io_struct.py (20250909) (#10236) · bf72b801
  Lianmin Zheng authored Sep 09, 2025
```
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: cctry <shiyang@x.ai>
```
  bf72b801
- [HiCache] feat: add mooncake backend extra config (#10213) · 8471e5e6
  Teng Ma authored Sep 10, 2025
  
  8471e5e6
- Revert "Revert the changes on NCCL symmetric memory" (#10238) · 4582931a
  Lianmin Zheng authored Sep 09, 2025
  
  4582931a
- Revert the changes on NCCL symmetric memory (#10210) · d352c29a
  Lianmin Zheng authored Sep 09, 2025
```
Co-authored-by: Yineng Zhang <me@zhyncs.com>
```
  d352c29a
- chore: upgrade v0.3.9 sgl-kernel (#10220) · d3ee7098
  Yineng Zhang authored Sep 09, 2025
  
  d3ee7098
- [Fix] KV-cache eviction mismatch across PP ranks in DeepSeek V3/R1 (#10214) · 71fc7b7f
  Rain H authored Sep 09, 2025
  
  71fc7b7f
- add variable TP Decode > Prefill size support (#9960) · 9ab72f98
  shaharmor98 authored Sep 09, 2025
```
Signed-off-by: Shahar Mor <smor@nvidia.com>
```
  9ab72f98
- [Auto Sync] Update sampling_batch_info.py (20250909) (#10212) · 71133a04
  Lianmin Zheng authored Sep 09, 2025
```
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: cctry <shiyang@x.ai>
```
  71133a04
- Refactor fused_add_rmsnorm import logic (#10207) · f5f6b3b4
  Shangming Cai authored Sep 09, 2025
```
Signed-off-by: Shangming Cai <csmthu@gmail.com>
```
  f5f6b3b4