Commits · 729f612dc69eb6f540e38f23e1b41f854b58a55e · change / sglang · GitLab

28 Oct, 2025 6 commits
- Update openai package version to 2.6.1 (#12222) · 729f612d
  Xinyuan Tong authored Oct 27, 2025
  
  729f612d
- Use explicit uint64 dtype for Tensor data_ptr() to avoid overflow (#11994) · 899453ac
  jianan-gu authored Oct 28, 2025
  
  899453ac
- Add env var to control custom Triton kernel cache and set CSGMV as default backend. (#12176) · ce832d70
  Lifu Huang authored Oct 27, 2025
  
  ce832d70
- Support running FP4 Deepseek on SM120. (#11708) · 88596739
  weiliang authored Oct 28, 2025
  
  88596739
- [Auto Sync] Update scheduler.py, spec_info.py, run_suite.py... (20251027) (#12235) · a6ea3add
  Yineng Zhang authored Oct 27, 2025
```
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: gongwei-130 <56567052+gongwei-130@users.noreply.github.com>
```
  a6ea3add
- Compiling rope while preserving true on policy (#12161) · 326c84c4
  fzyzcjy authored Oct 28, 2025
  
  326c84c4
27 Oct, 2025 13 commits
- fix: AttributeError: 'NixlKVManager' object has no attribute 'prefill_tp_size_table' (#12234) · 8da608cc
  gongwei-130 authored Oct 27, 2025
  
  8da608cc
- Add support for Matryoshka embeddings (#126) (#11142) · 9fc3e8aa
  satyamk7054 authored Oct 27, 2025
```
Co-authored-by: Satyam Kumar <satyamk@linkedin.com>
```
  9fc3e8aa
- rope xpu: fix missing argument 'fused_set_kv_buffer_arg' and replace native... · c11b34d5
  Chunyuan WU authored Oct 28, 2025
```
rope xpu: fix missing argument 'fused_set_kv_buffer_arg' and replace native with sgl_kernel_xpu impl (#12006)
```
  c11b34d5
- [Feature] PD-Multiplexing Context and Scheduler. (#11592) · 05ad28f2
  ykcombat authored Oct 28, 2025
  
  05ad28f2
- check_offload_progress more frequently (#11656) · 0cae873f
  pansicheng authored Oct 28, 2025
  
  0cae873f
- improve mimax-m2 rmsnorm precision (#12186) · a8b91f6b
  Haichao Zhu authored Oct 28, 2025
  
  a8b91f6b
- fix(metrics): double times add_latency for DECODE_BOOTSTRAP (#12209) · 959d1ab8
  Jimmy authored Oct 27, 2025
  
  959d1ab8
- [Detokenizer Manager] Cleanup state when reqs are finished (#12205) · 6c1c1933
  Muqi Li authored Oct 27, 2025
  
  6c1c1933
- Fix crash after flush cache (#12107) · 3029d301
  cctry authored Oct 27, 2025
```
Co-authored-by: fzyzcjy <5236035+fzyzcjy@users.noreply.github.com>
```
  3029d301
- Optimize triton_mrope with torch compile (#12112) · f389f017
  Yuan Luo authored Oct 27, 2025
```
Co-authored-by: luoyuan.luo <luoyuan.luo@antgroup.com>
```
  f389f017
- Add support for AutoRound quantized models (#10153) · caa4819b
  Weiwei authored Oct 27, 2025
  
  caa4819b
- GLM-4-0414 and GLM-4.1V Code Refactor (#12117) · a88b006e
  Yuxuan Zhang authored Oct 27, 2025
  
  a88b006e
- chore: bump SGLang version to 0.5.4.post1 (#12169) · 55d75e11
  sglang-bot authored Oct 27, 2025
  
  55d75e11
26 Oct, 2025 14 commits
- [misc][grpc] Remove duplicate log (#12168) · 94aad0de
  Chang Su authored Oct 26, 2025
  
  94aad0de
- [WIP] support MiniMax M2 model (#12129) · 7ebc28f5
  赵晨阳 authored Oct 26, 2025
```
Signed-off-by: Xinyuan Tong <xinyuantong.cs@gmail.com>
Signed-off-by: xuebi <xuebi@minimaxi.com>
Co-authored-by: Xinyuan Tong <xinyuantong.cs@gmail.com>
Co-authored-by: Xinyuan Tong <115166877+JustinTong0323@users.noreply.github.com>
Co-authored-by: Roger Young <42564206+rogeryoungh@users.noreply.github.com>
Co-authored-by: xuebi <xuebi@minimaxi.com>
```
  7ebc28f5
- transfer mrope_position_delta to device when first running (#11047) · 0b3b3e9a
  ash-sigh authored Oct 27, 2025
  
  0b3b3e9a
- Avoid using flashinfer_allreduce_fusion when dp attention is enabled. (#11632) · a1d5bc4c
  Elfie Guo authored Oct 26, 2025
  
  a1d5bc4c
- model: support NVILA and NVILA Lite (#10399) · a8023891
  Zijian Zhang authored Oct 27, 2025
  
  a8023891
- Support DeepGEMM for deterministic inference (#12142) · 0103f374
  fzyzcjy authored Oct 26, 2025
  
  0103f374
- [Fix] fix allreduce bug in Piecewise Graph (#12106) · 96a5a949
  zyksir authored Oct 26, 2025
  
  96a5a949
- Fix ITL metrics when using openai endpoint with spec (#12156) · ea385ae8
  Liangsheng Yin authored Oct 26, 2025
  
  ea385ae8
- [quantization] AWQ Marlin doesn't work when dtype is bfloat16 (#11494) · 6371f7af
  Kai-Hsun Chen authored Oct 26, 2025
```
Signed-off-by: Kai-Hsun Chen <khchen@x.ai>
Co-authored-by: Xinyuan Tong <115166877+JustinTong0323@users.noreply.github.com>
```
  6371f7af
- [misc] depdencies & enviroment flag (#12113) · 8491c794
  Liangsheng Yin authored Oct 26, 2025
  
  8491c794
- [log] Make forward iter count optional (#12116) · bda3758f
  Liangsheng Yin authored Oct 26, 2025
  
  bda3758f
- Clean up attention backend selection code & Other minor rename (#12136) · 7b36c47b
  Lianmin Zheng authored Oct 25, 2025
  
  7b36c47b
- [NVIDIA] Change default quant method for model_opt (#11991) · ff604064
  Kaixi Hou authored Oct 25, 2025
  
  ff604064
- Make bmm batch invariant injection optional (#12118) · c001deba
  fzyzcjy authored Oct 26, 2025
  
  c001deba
25 Oct, 2025 7 commits
- Clean up server launch code and multi tokenizer (#12132) · 8e70064c
  Lianmin Zheng authored Oct 25, 2025
  
  8e70064c
- Update sgl-kernel version to 0.3.16.post4 (#12125) · 4b0ac1d5
  Baizhou Zhang authored Oct 25, 2025
  
  4b0ac1d5
- Fix Illegal Instruction/IMA errors when using DP attention --... · c8492978
  YAMY authored Oct 25, 2025
```
Fix Illegal Instruction/IMA errors when using DP attention -- num_tokens_for_logprob calculation (#12115)
```
  c8492978
- Clean up server args & Add CI scripts (#12124) · 4caca1ba
  Lianmin Zheng authored Oct 25, 2025
  
  4caca1ba
- [Auto Sync] Update test_deterministic.py, test_deterministi... (20251024) (#12083) · ea13cb14
  Lianmin Zheng authored Oct 24, 2025
```
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Stefan He <hebiaobuaa@gmail.com>
```
  ea13cb14
- Fix 'KeyError' for per_token expert distribution recorder (#9501) · 89824189
  vipwangerxiao authored Oct 25, 2025
```
Signed-off-by: Peng Wang <rocking@linux.alibaba.com>
Co-authored-by: Peng Wang <rocking@linux.alibaba.com>
```
  89824189
- Support true on-policy (#12058) · 20bd2271
  fzyzcjy authored Oct 25, 2025
  
  20bd2271