Commits · 646076b71e8d12816d94de44a63ea57a568651de · change / sglang

"vscode:/vscode.git/clone" did not exist on "e4ae66868db048b968318970fa48cd670ed49705"

30 Aug, 2025 19 commits
- Update guidelines for syncing code between repos (#9831) · 646076b7
  Lianmin Zheng authored Aug 30, 2025
  
  646076b7
- [CI] Code sync tools (#9830) · 0d040089
  Lianmin Zheng authored Aug 30, 2025
  
  0d040089
- [CI] Fix the trigger condition for PR test workflows (#9761) · 05e47872
  Lianmin Zheng authored Aug 30, 2025
  
  05e47872
- [Auto Sync] Update parallel_state.py (20250830) (#9828) · 1e61b496
  Lianmin Zheng authored Aug 30, 2025
```
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
```
  1e61b496
- chore: upgrade transformers 4.56.0 (#9827) · 300676af
  Yineng Zhang authored Aug 30, 2025
  
  300676af
- [sgl-kernel] fix: fix missing FetchContent_Populate for fmt (#9826) · 7fe89f7c
  PGFLMG authored Aug 31, 2025
  
  7fe89f7c
- chore: upgrade sgl-kernel 0.3.7.post1 with deepgemm fix (#9822) · 9970e3bf
  Yineng Zhang authored Aug 30, 2025
  
  9970e3bf
- Fix typo in warning message about DeepGEMM JIT (#9802) · 70eedb58
  Mohammad Miadh Angkad authored Aug 30, 2025
  
  70eedb58
- chore: update Dockerfile (#9820) · 9c99949e
  Yineng Zhang authored Aug 30, 2025
  
  9c99949e
- chore: fix cuda driver api issue and bump sgl-kernel 0.3.7.post1 (#9746) · c5082f0f
  Yineng Zhang authored Aug 30, 2025
  
  c5082f0f
- Fix memory leak when aborting decode request in PD-Disagg (#9817) · 836873b9
  Liangsheng Yin authored Aug 30, 2025
```
Co-authored-by: Lianmin Zheng <15100009+merrymercy@users.noreply.github.com>
```
  836873b9
- fix: dsv3 lite q_lora_rank none (#9815) · 8abe8dea
  Yineng Zhang authored Aug 29, 2025
  
  8abe8dea
- Make fp4_quantize kernels work on sm103 (#9807) · 1e85589d
  hlu1 authored Aug 29, 2025
```
Signed-off-by: Hao Lu <14827759+hlu1@users.noreply.github.com>
```
  1e85589d
- feature(eplb): add min-rebalancing-utilization-threshold for eplb (#8345) · c2a26e72
  hzh0425 authored Aug 30, 2025
```
Co-authored-by: yizhang2077 <1109276519@qq.com>
```
  c2a26e72
- Small bug fix in transformers model implementation (#9809) · 591e6c59
  yilian49 authored Aug 29, 2025
  
  591e6c59
- Adds initialize_moe_config to bench_one_batch so MOE backend is respected (#9670) · 42f34437
  pranavm-nvidia authored Aug 29, 2025
  
  42f34437
- [NVIDIA] [2/N] Optimize `silu_and_mul_scaled_fp4_grouped_quant` perf (#9556) · 5c34b4f1
  Kaixi Hou authored Aug 29, 2025
  
  5c34b4f1
- Fix TRTLLM MLA Cuda KV Blocks Causing accuracy drop (#9675) · ff9b5618
  Faraz authored Aug 29, 2025
  
  ff9b5618
- [ModelOpt] Fix Weight Loading for DSR1-FP4 Quantization (#9712) · fcd72bd1
  Pavani Majety authored Aug 29, 2025
```
Signed-off-by: Pavani Majety <pmajety@nvidia.com>
```
  fcd72bd1
29 Aug, 2025 17 commits
- chore: upgrade flashinfer 0.3.0rc1 (#9793) · 3d8fc434
  Yineng Zhang authored Aug 29, 2025
  
  3d8fc434
- [feat] Support EAGLE3 for Qwen2 (#9216) · 87a0f7d2
  KerwinKai authored Aug 30, 2025
  
  87a0f7d2
- feat: add original logprobs to response (#8375) · 839c93bd
  narutolhy authored Aug 29, 2025
```
Co-authored-by: Chayenne <zhaochen20@outlook.com>
Co-authored-by: luhongyu.4869 <luhongyu.4869@bytedance.com>
```
  839c93bd
- feat: Add flexible validation for partial weight updates (#9663) · f1e9bbaf
  JiLi authored Aug 30, 2025
```
Co-authored-by: RichardW <rich-junwang@users.noreply.github.com>
Co-authored-by: Zhuorany <yzr1914001753@gmail.com>
Co-authored-by: Stefan He <hebiaobuaa@gmail.com>
Co-authored-by: Yineng Zhang <me@zhyncs.com>
Co-authored-by: Night <32424487+PrinsYin@users.noreply.github.com>
Co-authored-by: zhaochenyang20 <zhaochen20@outlook.com>
Co-authored-by: Liangsheng Yin <hnyls2002@gmail.com>
```
  f1e9bbaf
- support enable in the reasoning field to enable thingking for thinkin… (#9715) · 3fd1431d
  gongwei-130 authored Aug 29, 2025
  
  3fd1431d
- feat(hicache-3fs): 3FS-Store Backup Optimizations For MLA Model. (#9692) · 161e9dc5
  hzh0425 authored Aug 30, 2025
  
  161e9dc5
- [HiCache] resolve conflict between chunked-prefill and hicache hit count (#9776) · 54e872d3
  Zhiqiang Xie authored Aug 29, 2025
  
  54e872d3
- [PD] Support get_model_info interface for mini_lb (#9792) · e5b29bf1
  Xuchun Shang authored Aug 29, 2025
```
Signed-off-by: Xuchun Shang <xuchun.shang@linux.alibaba.com>
Co-authored-by: Teng Ma <sima.mt@alibaba-inc.com>
```
  e5b29bf1
- accomendate json schema in the "schema" field, not in "json_schema" field of... · 9a7c8842
  gongwei-130 authored Aug 28, 2025
```
accomendate json schema in the "schema" field, not in "json_schema" field of response_format (#9786)
```
  9a7c8842
- Make sm100 fp8 kernels available on sm103 (#9789) · 7a16db9b
  hlu1 authored Aug 28, 2025
```
Signed-off-by: Hao Lu <14827759+hlu1@users.noreply.github.com>
```
  7a16db9b
- add bench_mix.py (#9788) · 09a1df22
  pansicheng authored Aug 29, 2025
  
  09a1df22
- ROCm 7.0 update (#9757) · 4b7034dd
  sogalin authored Aug 28, 2025
  
  4b7034dd
- Raise error when `topk>1` and `page>1` for paged attention backends. (#9784) · a23c3020
  Liangsheng Yin authored Aug 29, 2025
```
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
```
  a23c3020
- Skip some tests on Blackwell (#9777) · a7d825fc
  hlu1 authored Aug 28, 2025
```
Signed-off-by: Hao Lu <14827759+hlu1@users.noreply.github.com>
```
  a7d825fc
- bugfix(hicache): Move exists check before key suffixing (#9749) · 38cd5fb1
  hzh0425 authored Aug 29, 2025
  
  38cd5fb1
- [HiCache] change the default policy to write through (#9772) · 001f5194
  Zhiqiang Xie authored Aug 28, 2025
  
  001f5194
- Optimize prefill performance on cpu backend (#8750) · 5ad296bd
  Ma Mingfei authored Aug 29, 2025
  
  5ad296bd
28 Aug, 2025 4 commits
- fix: fix MLA for ShardedModelLoader/RemoteModelLoader (#6287) · 9f81d741
  wangyu authored Aug 29, 2025
```
Signed-off-by: wangyu <wangyu.steph@bytedance.com>
```
  9f81d741
- feat(draft_model): support draft_model for RemoteModelLoader (#6407) · a38c1497
  wangyu authored Aug 29, 2025
```
Signed-off-by: wangyu <wangyu.steph@bytedance.com>
```
  a38c1497
- [Feature] Support NPUGraph for DeepSeek on Ascend NPU (#9355) · 74dd4249
  chenxu140 authored Aug 29, 2025
```
Co-authored-by: Even Zhou <even.y.zhou@outlook.com>
```
  74dd4249
- feat: add tuned fused moe config for GLM-4.5-Air-FP8 tp = 4 on B200 (#9770) · dc20c22f
  zixuanzhang226 authored Aug 28, 2025
  
  dc20c22f