Commits · b8b6008f47de97f2eecde78db1375aeeb63958ff · change / sglang

"vscode:/vscode.git/clone" did not exist on "1fd511e661ddf1a328ecbaa227009a58de7572f7"

03 Apr, 2025 10 commits
- [Fix] fix fa3 build at cu118 (#5036) · b8b6008f
  yinfan98 authored Apr 04, 2025
  
  b8b6008f
- Small refactor DeepEPMode to clean up code a bit (#4992) · 8e10fec9
  fzyzcjy authored Apr 03, 2025
  
  8e10fec9
- Replace enable_flashinfer_mla argument with attention_backend (#5005) · e8999b13
  Baizhou Zhang authored Apr 03, 2025
  
  e8999b13
- try to fix ci oserror (#5024) · 772d2a19
  Xiaoyu Zhang authored Apr 03, 2025
  
  772d2a19
- fix deepgemm as well (#5030) · 9d0b36c4
  Zhiqiang Xie authored Apr 03, 2025
  
  9d0b36c4
- [Build] Support build sgl-kernel with ccache (#5020) · 7d8c0ce7
  Yuhong Guo authored Apr 03, 2025
  
  7d8c0ce7
- fix: fix illegal cuda memory access at fused_moe_kernel (#4727) · e41549c3
  saltyfish66 authored Apr 03, 2025
```
Co-authored-by: yuethe <yuethe@tencent.com>
```
  e41549c3
- Feature/revise docs ci (#5009) · cccfc10e
  renxin authored Apr 02, 2025
  
  cccfc10e
- update cutlass tag (#5011) · a2aea59b
  Zhiqiang Xie authored Apr 02, 2025
  
  a2aea59b
- [sgl-kernel] per token group quant support COLUMN MAJOR (#4817) · 2c8fd993
  Xiaoyu Zhang authored Apr 03, 2025
  
  2c8fd993
02 Apr, 2025 6 commits
- Update tokenizer_manager.py (#5008) · 31da75ab
  Kaiyu Yang authored Apr 02, 2025
  
  31da75ab
- Add Eagle Speculative Decoding to FA3 Backend (#4951) · e983e432
  Qingquan Song authored Apr 02, 2025
```
Co-authored-by: hebiao064 <hebiaobuaa@gmail.com>
Co-authored-by: Baizhou Zhang <sobereddiezhang@gmail.com>
Co-authored-by: zcnrex <zcnrex@gmail.com>
```
  e983e432
- sgl scaled_fp8_quant support output padding (#4861) · e9c6ce46
  Xiaoyu Zhang authored Apr 02, 2025
  
  e9c6ce46
- bug fix for hicache host eviction (#4989) · 3fadc647
  Zhiqiang Xie authored Apr 02, 2025
  
  3fadc647
- Large page size aligned hierarchical caching (#4581) · e119f042
  Zhiqiang Xie authored Apr 01, 2025
  
  e119f042
- [VLM RLHF] Take Image input for verl vlm rollout (#4915) · 9eb49e87
  XinyuanTong authored Apr 01, 2025
```
Signed-off-by: Xinyuan Tong <justinning0323@outlook.com>
Co-authored-by: GeLee <leege233@gmail.com>
```
  9eb49e87
01 Apr, 2025 4 commits
- Prevent memory leak of retract_decode when page_size > 1 (#4977) · 12047f5e
  Zhiqiang Xie authored Apr 01, 2025
  
  12047f5e
- update bench_serving (#4958) · fda6bb78
  Yineng Zhang authored Apr 01, 2025
  
  fda6bb78
- [Feature] Support DeepEP Low Latency (#4767) · 23c764b1
  Jinyan Chen authored Apr 02, 2025
```
Co-authored-by: sleepcoo <sleepcoo@gmail.com>
Co-authored-by: laixinn <xielx@shanghaitech.edu.cn>
Co-authored-by: ch-wan <cwan39@gatech.edu>
```
  23c764b1
- Revert PR 4764 & 4813 related to R1 RoPE (#4959) · 87fafa01
  Yuhong Guo authored Apr 01, 2025
  
  87fafa01
31 Mar, 2025 10 commits
- use fa3 in sgl-kernel (#4954) · 1c63e797
  Yineng Zhang authored Mar 31, 2025
  
  1c63e797
- [Build] Fix cuda12.8 build error in nvfp4_scaled_mm_kernels.cu (#4953) · ee47a6c1
  Yuhong Guo authored Apr 01, 2025
  
  ee47a6c1
- bump sgl-kernel v0.0.6 (#4950) · 6384d317
  Yineng Zhang authored Mar 31, 2025
  
  6384d317
- refactor: multimodal data (#4754) · 5cb552b1
  Mick authored Apr 01, 2025
  
  5cb552b1
- [Fix] revert clean m.def for cudagraph (#4944) · c7457191
  yinfan98 authored Mar 31, 2025
  
  c7457191
- [feat] interface for platforms abstraction (#4928) · 51ac297a
  JieXin Liang authored Mar 31, 2025
  
  51ac297a
- Fix oom error for large page size (#4913) · a169b9f8
  Zhiqiang Xie authored Mar 30, 2025
```
Co-authored-by: Lianmin Zheng <lianminzheng@gmail.com>
```
  a169b9f8
- [Fix] Add torch compile for torch.clamp back (#4936) · 4a63bc32
  Baizhou Zhang authored Mar 30, 2025
  
  4a63bc32
- Fix DeepSeek bug causing 2.2% MMLU drop when TP!=DP (#4883) · a303325f
  fzyzcjy authored Mar 31, 2025
```
Co-authored-by: ch-wan <cwan39@gatech.edu>
```
  a303325f
- [Fix] Improve Lora tests and reduce CI runtime (#4925) · 42873eac
  Baizhou Zhang authored Mar 30, 2025
  
  42873eac
30 Mar, 2025 10 commits
- cleanup sgl-kernel (#4933) · 4814ecaf
  Yineng Zhang authored Mar 30, 2025
  
  4814ecaf
- [Fix] avoid stream sync and torch compile in prefill for fa3 backend (#4932) · e62d60fe
  Baizhou Zhang authored Mar 30, 2025
  
  e62d60fe
- Fix sglang frontend's incorrect dependency on torch (#4931) · 032f8faa
  SEPLOS authored Mar 31, 2025
  
  032f8faa
- [feat] add fa3 in sgl-kernel (#4902) · 37c66ec8
  yinfan98 authored Mar 31, 2025
```
Co-authored-by: Sleepcoo <Sleepcoo@gmail.com>
```
  37c66ec8
- Fix 2-gpu CI test and suppress some warnings (#4930) · 9adf178c
  Lianmin Zheng authored Mar 30, 2025
  
  9adf178c
- Fix the timeout for unit-test-2-gpu in pr-test.yml (#4927) · f842853a
  Lianmin Zheng authored Mar 30, 2025
  
  f842853a
- fix bmm fp8 (#4926) · 195a09f5
  Yineng Zhang authored Mar 30, 2025
  
  195a09f5
- [Feature] use pytest for sgl-kernel (#4896) · 9fccda31
  Adarsh Shirawalmath authored Mar 30, 2025
  
  9fccda31
- Fix retract for page size > 1 (#4914) · 4ede6770
  Lianmin Zheng authored Mar 30, 2025
  
  4ede6770
- Support page size > 1 + eagle (#4908) · b26bc86b
  Lianmin Zheng authored Mar 30, 2025
  
  b26bc86b