Commits · 2695ab05375456b4a2ac625cca12b95c3b90e5ae · change / sglang

"git@developer.sourcefind.cn:wangsen/mineru.git" did not exist on "99cf160d1cbe475b9a7e0ce4ae9af935c14e7f89"

08 Apr, 2025 10 commits
- Fix loading KV quantization scale; Enable modelopt kv cache (#4686) · 2695ab05
  Yun Dai authored Apr 08, 2025
```
Co-authored-by: qingquansong <ustcsqq@gmail.com>
```
  2695ab05
- Fix torch compile errors (#5158) · 88d6fd9a
  kk authored Apr 08, 2025
  
  88d6fd9a
- fix empty_cache error in pt_weights_iterator (#5151) · cc88d98a
  DangKai authored Apr 08, 2025
```
Co-authored-by: dangkai.dk <dangkai.dk@alibaba-inc.com>
```
  cc88d98a
- Add unit test on page_size > 1 and mla and integration test for Flash Attention 3 (#4760) · 804d9f2e
  Yubo Wang authored Apr 07, 2025
  
  804d9f2e
- [FA3 Feature] Support multi modal Llama-3.2-11B-Vision-Instruct (#5103) · a7c3f74b
  Chunan Zeng authored Apr 07, 2025
  
  a7c3f74b
- Fix run time error in ROCm platform (#5147) · 5a144a8a
  kk authored Apr 08, 2025
```
Co-authored-by: wunhuang <wunhuang@amd.com>
Co-authored-by: root <root@dell300x-pla-t10-17.pla.dcgpu>
```
  5a144a8a
- fix multimodal hash feature (#5083) · 27f8e6b9
  huangtingwei authored Apr 08, 2025
  
  27f8e6b9
- [AMD] Fix missing per_token_group_quant_fp8 for ROCm (#5140) · afb752bc
  Hubert Lu authored Apr 07, 2025
  
  afb752bc
- [modelopt] automatically inspect if model is ModelOpt quantized and set quantization method (#5145) · 9731eca7
  Yun Dai authored Apr 07, 2025
  
  9731eca7
- feat: disable grammar restrictions within reasoning sections (#4984) · 7c5658c1
  mlmz authored Apr 08, 2025
```
Co-authored-by: tianhaoyu <thy@mail.ecust.edu.cn>
Co-authored-by: DarkSharpness <2040703891@qq.com>
```
  7c5658c1
07 Apr, 2025 5 commits
- Refactor and Optimize FA3 Code (#5090) · 93470a14
  Stefan He authored Apr 07, 2025
```
Co-authored-by: Qingquan Song <ustcsqq@gmail.com>
```
  93470a14
- [ci] fix llama4 ci error (#5126) · db452760
  Xiaoyu Zhang authored Apr 07, 2025
  
  db452760
- bump v0.4.5 (#5117) · 57f99608
  Yineng Zhang authored Apr 07, 2025
  
  57f99608
- Fix refactor error - fp8.py (#5106) · 81992474
  HAI authored Apr 07, 2025
```
Co-authored-by: Lianmin Zheng <lianminzheng@gmail.com>
```
  81992474
- Add Llama4 support (#5092) · f04c80dc
  Chang Su authored Apr 07, 2025
```
Co-authored-by: Cheng Wan <cwan39@gatech.edu>
Co-authored-by: fzyzcjy <ch271828n@outlook.com>
Co-authored-by: ispobock <ispobaoke@163.com>
```
  f04c80dc
05 Apr, 2025 5 commits
- bump v0.4.4.post4 (#5091) · 35e0856b
  Yineng Zhang authored Apr 05, 2025
  
  35e0856b
- python transfer custom allreduce from trt kernel to vllm kernel (#5080) · aba5ca15
  Yi Zhang authored Apr 06, 2025
  
  aba5ca15
- upgrade transformers 4.51.0 (#5088) · 0d99adb7
  Yineng Zhang authored Apr 05, 2025
  
  0d99adb7
- [Revision] Replace enable_flashinfer_mla argument with attention_backend (#5052) · efbae697
  Baizhou Zhang authored Apr 05, 2025
  
  efbae697
- FA3 Spec Decoding to support top k = 1 and add cuda graph support (#5050) · ca8d02ab
  Stefan He authored Apr 04, 2025
```
Co-authored-by: Qingquan Song <ustcsqq@gmail.com>
Co-authored-by: Chunan Zeng <zcnrex@gmail.com>
```
  ca8d02ab
04 Apr, 2025 11 commits
- fix dummy-load deepseekv2 (#4535) · 7ed77d6b
  inkcherry authored Apr 05, 2025
  
  7ed77d6b
- [deepep] fix: shared experts are not initialized when shared experts fusion is enabled (#5072) · 4c54f442
  Cheng Wan authored Apr 04, 2025
  
  4c54f442
- Add DeepSeek V3/R1 shared experts fusion (#4918) · 924ca7c9
  Xiaoyu Zhang authored Apr 04, 2025
  
  924ca7c9
- Cleanup unused resources after DeepEP operation (#4996) · 6ff9c6a5
  fzyzcjy authored Apr 04, 2025
  
  6ff9c6a5
- Support async DeepEP by splitting into two stages (#4995) · 77e929a1
  fzyzcjy authored Apr 04, 2025
  
  77e929a1
- Small refactor DeepEPDispatcher into subclasses (#4994) · febe21ce
  fzyzcjy authored Apr 04, 2025
  
  febe21ce
- [fix] remove `cuda_device_count_stateless` (#5060) · a995a773
  JieXin Liang authored Apr 04, 2025
  
  a995a773
- Add H20 fused MoE kernel tuning configs for DeepSeek V3/R1 (#5057) · 31035dda
  Tommy Yang authored Apr 04, 2025
  
  31035dda
- [2/3] fix dsv3 awq issue (#4625) · d95269f9
  AniZpZ authored Apr 04, 2025
```
Co-authored-by: 晟海 <huangtingwei.htw@antgroup.com>
Co-authored-by: laixinn <xielx@shanghaitech.edu.cn>
```
  d95269f9
- upgrade sgl-kernel v0.0.7 (#5049) · e53bf190
  Yineng Zhang authored Apr 03, 2025
  
  e53bf190
- Update the retry count (#5051) · 3289c120
  Yineng Zhang authored Apr 03, 2025
  
  3289c120
03 Apr, 2025 5 commits
- Add LlavaLlamaForCausaLM in MultiModal Processors (#5039) · 69df9761
  Ravi Theja authored Apr 04, 2025
```
Co-authored-by: Ravi Theja Desetty <ravitheja@Ravis-MacBook-Pro.local>
```
  69df9761
- Revert "Replace enable_flashinfer_mla argument with attention_backend" (#5048) · 74885a84
  Lianmin Zheng authored Apr 03, 2025
  
  74885a84
- 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
- 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
02 Apr, 2025 4 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