Commits · 80b2b3207a086061f419b7ae38ccec7dc5562f69 · change / sglang · GitLab

22 Oct, 2025 4 commits
- Enable native ModelOpt quantization support (3/3) (#10154) · 80b2b320
  Zhiyu authored Oct 21, 2025
```
Signed-off-by: Zhiyu Cheng <zhiyuc@nvidia.com>
```
  80b2b320
- [lint] improve ruff check (#11922) · 9d61205d
  Liangsheng Yin authored Oct 22, 2025
```
Co-authored-by: Xiaoyu Zhang <35585791+BBuf@users.noreply.github.com>
```
  9d61205d
- [router][grpc] Support `v1/responses` API (#11926) · 70f6309c
  Chang Su authored Oct 21, 2025
  
  70f6309c
- fix: resolve flashinfer 0.4.1 import (#11940) · 70416001
  Yineng Zhang authored Oct 21, 2025
  
  70416001
21 Oct, 2025 23 commits
- [Auto Sync] Update forward_batch_info.py (20251021) (#11934) · c461e771
  Yineng Zhang authored Oct 21, 2025
```
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: yinghui <32845984+cicirori@users.noreply.github.com>
```
  c461e771
- [BugFix][Qwen3-VL]: add metadata for video in qwen3-vl (#11377) · fde2decf
  Zheng Wengang authored Oct 22, 2025
  
  fde2decf
- chore: upgrade flashinfer 0.4.1 (#11933) · 9792b9d7
  Yineng Zhang authored Oct 21, 2025
  
  9792b9d7
- Rename flashmla kernel options of nsa backend for better readability (#11876) · ef4a8097
  Baizhou Zhang authored Oct 21, 2025
  
  ef4a8097
- Update sgl-kernel and remove fast hadamard depedency (#11844) · ebff4ee6
  Baizhou Zhang authored Oct 21, 2025
  
  ebff4ee6
- [NVIDIA] Add new SMs support for Spark & Thor (#11287) · 2b1da821
  Serge Panev authored Oct 21, 2025
```
Signed-off-by: Serge Panev <spanev@nvidia.com>
```
  2b1da821
- Fix flush cache API for spec v2 (#11918) · 97710ccd
  Liangsheng Yin authored Oct 21, 2025
  
  97710ccd
- [quantization][MoE] fix the check for `tp_size` / `moe_ep_size` /... · c61b0b29
  Kai-Hsun Chen authored Oct 21, 2025
```
[quantization][MoE] fix the check for `tp_size` / `moe_ep_size` / `moe_intermediate_size` / `weight_block_size_n` (#11702)
Signed-off-by: Kai-Hsun Chen <khchen@x.ai>
```
  c61b0b29
- [smol] [perf] Inverse perm improvement (#11482) · e8640ee9
  Vincent Zhong authored Oct 21, 2025
```
Signed-off-by: vincentzed <207368749+vincentzed@users.noreply.github.com>
```
  e8640ee9
- vlm: enforce pybase64 for image and str encode/decode (#10700) · d0a64c7e
  b8zhong authored Oct 21, 2025
  
  d0a64c7e
- Simplify multi-tokenizer (#11295) · 260fe755
  Zhengke Zhou authored Oct 21, 2025
```
Signed-off-by: zhengkezhou1 <madzhou1@gmail.com>
Co-authored-by: Liangsheng Yin <lsyincs@gmail.com>
```
  260fe755
- Support Thinking Budget (via custom_logit_processor for OpenAI API) [Fix #6572] (#11416) · dbb16bed
  ybyang authored Oct 21, 2025
```
Signed-off-by: ybyang <ybyang7@iflytek.com>
Co-authored-by: YorkSu <york_su@qq.com>
```
  dbb16bed
- [FEATURE] Add OpenAI-Compatible LoRA Adapter Selection (#11570) · 852c0578
  Neelabh Sinha authored Oct 21, 2025
  
  852c0578
- init support for KTransformers Heterogeneous Computing (#11487) · 7e6191c0
  Atream authored Oct 21, 2025
```
Co-authored-by: Jianwei Dong <1913953267@qq.com>
```
  7e6191c0
- [AMD] Update wave-lang to 3.8.0 (#11878) · 6f9b66bd
  Gaurav Verma authored Oct 20, 2025
```
Signed-off-by: xintin <gaurav.verma@amd.com>
```
  6f9b66bd
- Use trtllm_mla decode kernel for draft extend in speculative decoding (#11664) · d9a20fd2
  Qiaolin Yu authored Oct 20, 2025
  
  d9a20fd2
- Init attention backend for Intel XPU (#10656) · b113c72e
  Meng, Hengyu authored Oct 21, 2025
```
Co-authored-by: guangyey <guangye.yu@intel.com>
Co-authored-by: DiweiSun <105627594+DiweiSun@users.noreply.github.com>
```
  b113c72e
- Fix RotaryEmbedding for fp32 input (#11843) · fb6cc7b0
  zhangdonghao-zdh authored Oct 21, 2025
  
  fb6cc7b0
- piecewise cuda graph support qwen3-moe (#11845) · 8374a96e
  Xiaoyu Zhang authored Oct 21, 2025
  
  8374a96e
- Revise MRotaryEmbedding's forward (#11859) · 74de76c6
  Yuan Luo authored Oct 21, 2025
```
Co-authored-by: luoyuan.luo <luoyuan.luo@antgroup.com>
Co-authored-by: 羽癫 <yudian.zy@antgroup.com>
Co-authored-by: b8zhong <b8zhong@uwaterloo.ca>
```
  74de76c6
- [router][grpc] Fix wram-up random token ids for small models (#11887) · 9c0b1eb5
  Chang Su authored Oct 20, 2025
  
  9c0b1eb5
- [code move] move pp into a separate mixin (#11838) · 01f14a7a
  Lianmin Zheng authored Oct 20, 2025
  
  01f14a7a
- [Auto Sync] Update scheduler.py, server_args.py (20251020) (#11875) · 43ad0590
  Lianmin Zheng authored Oct 20, 2025
```
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Kan Wu <wukanustc@gmail.com>
```
  43ad0590
20 Oct, 2025 13 commits
- Support mixing cutedsl and deepgemm backend (#11807) · 0917c5da
  fzyzcjy authored Oct 21, 2025
  
  0917c5da
- Replace function call with set literal (#11867) · 184a4df6
  penguin_wwy authored Oct 21, 2025
  
  184a4df6
- Fix acc len and gen throughput metrics when enabling overlap-spec (#11823) · f7b1d8c5
  Qiaolin Yu authored Oct 20, 2025
```
Co-authored-by: Liangsheng Yin <lsyincs@gmail.com>
```
  f7b1d8c5
- [9/N] MoE Refactor: cleanup dispatcher interfaces (#11847) · bfc3b3f7
  Cheng Wan authored Oct 20, 2025
  
  bfc3b3f7
- Tiny fix main lint (#11862) · da5bde4d
  Liangsheng Yin authored Oct 20, 2025
  
  da5bde4d
- [Feature] New structural tag support (#10691) · 276e7b3e
  DarkSharpness authored Oct 20, 2025
  
  276e7b3e
- fix(server_args): handle tokenizer init conflicts (#11776) · 296f6892
  ishandhanani authored Oct 20, 2025
  
  296f6892
- [Model] Add Olmo 3 model support (#11396) · d383e661
  Shane A authored Oct 19, 2025
  
  d383e661
- Tiny clean up for PD module and doc (#11747) · a2ba0bc3
  Shangming Cai authored Oct 20, 2025
```
Signed-off-by: Shangming Cai <csmthu@gmail.com>
```
  a2ba0bc3
- [PD] Improve eagle acceptance rate by transferring draft model hidden states (#10801) · 6d2d0ce2
  Ziming Huang authored Oct 20, 2025
```
Co-authored-by: Shangming Cai <csmthu@gmail.com>
```
  6d2d0ce2
- Support mrope triton kernel and add unit test (#11722) · 271d3d0d
  Yuan Luo authored Oct 20, 2025
```
Co-authored-by: luoyuan.luo <luoyuan.luo@antgroup.com>
Co-authored-by: b8zhong <b8zhong@uwaterloo.ca>
```
  271d3d0d
- [Feature] Use current greenctx stream to communicate in PD-Multiplexing. (#11594) · c4e81e64
  ykcombat authored Oct 20, 2025
  
  c4e81e64
- Recapture cuda graph after model weight update to resolve IMA error (#11780) · c726d44c
  harrisonlimh authored Oct 19, 2025
  
  c726d44c