Commits · 1b2ff4fb7f05ec82128765c366e6f75f4e3f05f7 · change / sglang · GitLab

03 Sep, 2025 11 commits
- Revert "Optimized deepseek-v3/r1 model performance on mxfp4 run (#9671)" (#9959) · 1b2ff4fb
  Yineng Zhang authored Sep 03, 2025
  
  1b2ff4fb
- Revert "[Fix] DeepSeek EP accuracy issue on B200 GPUs (#9946)" (#9955) · 2c7ca33a
  Yineng Zhang authored Sep 02, 2025
  
  2c7ca33a
- [feat] Add P/D attention select for draft model (#9755) · df397a72
  Ximingwang-09 authored Sep 03, 2025
```
Co-authored-by: 纬杭 <ximing.wxm@antgroup.com>
```
  df397a72
- add proctitle for tokenizers (#9952) · 5dfcd6c2
  Liangsheng Yin authored Sep 03, 2025
```
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
```
  5dfcd6c2
- Optimized deepseek-v3/r1 model performance on mxfp4 run (#9671) · 0dfd54d1
  kk authored Sep 03, 2025
```
Co-authored-by: wunhuang <wunhuang@amd.com>
Co-authored-by: wghuang <wghuang@amd.com>
```
  0dfd54d1
- Qwen FP8/NVFP4 ModelOPT Quantization support (#7912) · bcbeed71
  jingyu-ml authored Sep 02, 2025
```
Co-authored-by: Jingyu Xin <jingyux@nvidia.com>
```
  bcbeed71
- Update tool_chat_template_deepseekv31.jinja (#9895) · cc9a31c6
  南京小汤包 authored Sep 03, 2025
  
  cc9a31c6
- Remove annoying warnings in sgl kernel build (#9905) · d631290e
  Lianmin Zheng authored Sep 02, 2025
  
  d631290e
- fix(cache): move ongoing_prefetch pop after validation to prevent leak (#9927) · 37565b7f
  JinYan Su authored Sep 03, 2025
```
Co-authored-by: Zhiqiang Xie <xiezhq@stanford.edu>
```
  37565b7f
- [Fix] DeepSeek EP accuracy issue on B200 GPUs (#9946) · 6243c367
  Al-Ekram Elahee Hridoy authored Sep 02, 2025
  
  6243c367
- Move parsers under a single folder (#9912) · 60e37f80
  Lianmin Zheng authored Sep 02, 2025
  
  60e37f80
02 Sep, 2025 23 commits
- [HiCache] Minor fix on file storage backend (#9869) · 369b1433
  Zhiqiang Xie authored Sep 02, 2025
  
  369b1433
- [model] support MiniCPM-V 4.0 (#8747) · 03dbf1aa
  tc-mb authored Sep 03, 2025
```
Signed-off-by: tc-mb <caitianchi@modelbest.cn>
Co-authored-by: Xinyuan Tong <115166877+JustinTong0323@users.noreply.github.com>
```
  03dbf1aa
- Grpc client (#9939) · 11dcabc5
  Chang Su authored Sep 02, 2025
  
  11dcabc5
- Fix the key passing issue in page first layout. (#9929) · 4d89389c
  hzh0425 authored Sep 03, 2025
  
  4d89389c
- [router] include rust benchamrks (#9932) · 9491d6e5
  Simo Lin authored Sep 02, 2025
  
  9491d6e5
- Support the internvl3.5 family models in sglang (#9705) · f64b8e3e
  yilian49 authored Sep 02, 2025
  
  f64b8e3e
- [Hicache] Generic page get bugfix (#9909) · 53976fce
  ykwd authored Sep 02, 2025
  
  53976fce
- chore: bump v0.5.2rc1 (#9920) · 18f91eb6
  Yineng Zhang authored Sep 02, 2025
  
  18f91eb6
- fix: update router deps (#9921) · 8766b3ac
  Yineng Zhang authored Sep 02, 2025
  
  8766b3ac
- [feat] apply deep_gemm compile_mode to skip launch (#9879) · 1db649ac
  JieXin Liang authored Sep 02, 2025
  
  1db649ac
- fix parallel_state.py `current_platform` bug (#9919) · a1e5d781
  Xiaoyu Zhang authored Sep 02, 2025
  
  a1e5d781
- [Fix] fix the issue encountered when inference LongCat-Flash/MTP EP MoE on b200 (#9916) · b7361cc4
  Guoyuan Lin authored Sep 02, 2025
  
  b7361cc4
- chore: bump v0.3.8 sgl-kernel (#9907) · a96c5b5c
  Yineng Zhang authored Sep 02, 2025
  
  a96c5b5c
- Change tensor alignment method to mn major (#9844) · b9eb0d9c
  Mohammad Miadh Angkad authored Sep 02, 2025
  
  b9eb0d9c
- [chore] fix dead links in doc (#9913) · 1fbfdebe
  Lifu Huang authored Sep 02, 2025
  
  1fbfdebe
- Move multi-tokenizer event loop to better place (#9902) · a25e8e42
  Shangming Cai authored Sep 02, 2025
```
Signed-off-by: Shangming Cai <csmthu@gmail.com>
```
  a25e8e42
- [feat] Support tp mode for DeepSeek-R1-W4AFP8 (#8118) · d4a93841
  chenxj authored Sep 02, 2025
```
Co-authored-by: yuhyao <827623970@qq.com>
```
  d4a93841
- [router] fix FunctionCallResponse proto, support arguments is null (#9875) · 21e1bc47
  Bruce-x-1997 authored Sep 02, 2025
```
Co-authored-by: forestlee95 <forestlee95@foxmail.com>
```
  21e1bc47
- [router] add grpc pd and regular router init (#9893) · 9a0cac1b
  Chang Su authored Sep 01, 2025
  
  9a0cac1b
- [code style] restruct fused_moe to avoid very long single file (#9878) · b5245064
  Xiaoyu Zhang authored Sep 02, 2025
  
  b5245064
- [router] Fix short timeout for the prefill client (#9803) · 9d9fa9a5
  LukasBluebaum authored Sep 02, 2025
  
  9d9fa9a5
- [HiCacheStorage]: Improve 3fs kvstore‘s performance and resolve mla issues (#9876) · 58d06fdc
  hzh0425 authored Sep 02, 2025
  
  58d06fdc
- [HiCacheStorage] fix abort request host memory leaks (#9874) · cb9e0e41
  huangtingwei authored Sep 02, 2025
```
Co-authored-by: Zhiqiang Xie <xiezhq@stanford.edu>
```
  cb9e0e41
01 Sep, 2025 6 commits
- support fp8 kvcache for hybrid attn backend on GPT-OSS (#9783) · 9db80253
  Rain Jiang authored Sep 01, 2025
```
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
```
  9db80253
- [router] add tokenizer download support from hf hub (#9882) · 598c0bc1
  Chang Su authored Sep 01, 2025
  
  598c0bc1
- Mooncake store get zero copy meta optimization (#9857) · b361750a
  huangtingwei authored Sep 01, 2025
  
  b361750a
- chore: bump v0.5.2rc0 (#9862) · 16e56ea6
  Yineng Zhang authored Sep 01, 2025
  
  16e56ea6
- chore: upgrade flashinfer 0.3.0 (#9864) · 349b491c
  Yineng Zhang authored Sep 01, 2025
  
  349b491c
- Support Multi Process Tokenizer Manager(#6555) (#8964) · 5f77e129
  ybyang authored Sep 01, 2025
```
Signed-off-by: ybyang <ybyang7@iflytek.com>
Signed-off-by: huanglong <huanglong@linux.alibaba.com>
Co-authored-by: Huang Long <121648372+LLLL114@users.noreply.github.com>
Co-authored-by: huanglong <huanglong@linux.alibaba.com>
Co-authored-by: Shangming Cai <csmthu@gmail.com>
```
  5f77e129