Commits · ab1a4fa5cbb0cf5d2ea9a2f3b36f16fd129cf945 · change / sglang · GitLab

14 Jun, 2025 11 commits
- [fix] fix cutlass_mla_backend with cuda_graph and add sm_scale for sgl-kernel cutlass_mla (#7184) · ab1a4fa5
  JieXin Liang authored Jun 15, 2025
  
  ab1a4fa5
- [fix] fix dsv3 weight loader tqdm and simplify shared experts fusion (#7181) · ed54bf9d
  JieXin Liang authored Jun 15, 2025
  
  ed54bf9d
- Fix shared experts fusion + weight requant (#7177) · b57d87c2
  fzyzcjy authored Jun 14, 2025
  
  b57d87c2
- Add Phi-4-mm to supported VLM supported model list. (#7178) · 98538822
  Lifu Huang authored Jun 13, 2025
  
  98538822
- Increase timeout in test/srt/test_disaggregation.py (#7175) · f47a1b1d
  Lianmin Zheng authored Jun 13, 2025
  
  f47a1b1d
- Support new DeepGEMM (#7172) · 93cec433
  fzyzcjy authored Jun 14, 2025
  
  93cec433
- Improve test cases for eagle infer (#7173) · ba589b88
  Lianmin Zheng authored Jun 13, 2025
  
  ba589b88
- Add test for refactored openai server (#7161) · 50876abc
  Jinn authored Jun 13, 2025
  
  50876abc
- Refactor DeepGEMM integration (#7150) · b4c41f72
  fzyzcjy authored Jun 14, 2025
  
  b4c41f72
- Support new DeepGEMM input format in silu_and_mul_masked_post_quant_fwd (#7153) · 8b8f2e74
  fzyzcjy authored Jun 14, 2025
  
  8b8f2e74
- Split the eagle test into two files (#7170) · 0fc3d992
  Lianmin Zheng authored Jun 13, 2025
  
  0fc3d992
13 Jun, 2025 13 commits
- Minor style change of triton backend (#7165) · be2d985d
  Lianmin Zheng authored Jun 13, 2025
  
  be2d985d
- Re-quantize DeepSeek model weights to support DeepGEMM new input format (#7156) · 5b1afa78
  fzyzcjy authored Jun 14, 2025
  
  5b1afa78
- Remove 200us slow concat kernel (part 2: srt) (#7020) · c49c1d92
  fzyzcjy authored Jun 14, 2025
  
  c49c1d92
- Tiny add sanity checks for DeepGEMM inputs (#7157) · 0f1dfa1e
  fzyzcjy authored Jun 14, 2025
  
  0f1dfa1e
- Minor speed up block_quant_dequant (#6814) · e3ec6bf4
  fzyzcjy authored Jun 14, 2025
  
  e3ec6bf4
- Fix DeepEP error in some environments (#7154) · b04df75a
  fzyzcjy authored Jun 14, 2025
  
  b04df75a
- Support new DeepGEMM format in per token group quant (part 2: srt) (#7155) · bec3e484
  fzyzcjy authored Jun 14, 2025
  
  bec3e484
- chore: bump v0.1.8.post1 (#7152) · 8ab7d93c
  Yineng Zhang authored Jun 13, 2025
  
  8ab7d93c
- Support new DeepGEMM format in per token group quant (#7146) · 5c66c442
  fzyzcjy authored Jun 13, 2025
  
  5c66c442
- Remove 200us slow concat kernel (part 1: kernel) (#7145) · aa46ed34
  fzyzcjy authored Jun 13, 2025
  
  aa46ed34
- filter by num_hidden_layers (#7056) · 2f4ec752
  pansicheng authored Jun 13, 2025
  
  2f4ec752
- Minor speedup topk postprocessing (#7058) · da47621c
  fzyzcjy authored Jun 13, 2025
  
  da47621c
- Remove unnecessary metadata_expand.max_seq_len_k operations in fa3 to… (#7140) · 22a6b9fc
  Binyao Jiang authored Jun 12, 2025
  
  22a6b9fc
12 Jun, 2025 8 commits
- Minor add utility to read expert distribution recorder output (#7134) · b02df20a
  fzyzcjy authored Jun 13, 2025
  
  b02df20a
- [Fix] Reduce busy polling when scheduler is idle (#6026) · bd7cfbd2
  Povilas Kanapickas authored Jun 13, 2025
  
  bd7cfbd2
- Add gfx950 support for sgl-kernel. (#7092) · 4b9971e4
  sogalin authored Jun 13, 2025
```
Co-authored-by: HAI <hixiao@gmail.com>
Co-authored-by: Yineng Zhang <me@zhyncs.com>
```
  4b9971e4
- Update CODEOWNERS (#7126) · dcc79d32
  Lianmin Zheng authored Jun 12, 2025
  
  dcc79d32
- feat: update blackwell setup (#7119) · 7046e0fa
  Yineng Zhang authored Jun 12, 2025
  
  7046e0fa
- Improve log status (#7115) · 930746d9
  Liangsheng Yin authored Jun 12, 2025
  
  930746d9
- [sgl-kernel] Add cuda kernel for moe_ep_silu_and_mul (#6919) · 84727a51
  Yuan Luo authored Jun 12, 2025
```
Co-authored-by: luoyuan.luo <luoyuan.luo@antgroup.com>
```
  84727a51
- Fix positional argument (#7093) · ef326774
  Quanfeng Li authored Jun 12, 2025
  
  ef326774
11 Jun, 2025 8 commits
- [Perf] Refactor LoRAManager to eliminate stream syncs and redundant computations (#6994) · 021f76e4
  Lifu Huang authored Jun 11, 2025
  
  021f76e4
- [feat]: Emit fixed-size KV blocks events (#6824) · 777688b8
  Faradawn Yang authored Jun 11, 2025
  
  777688b8
- [FIX]remove redundant code in logits_processor.py (#7079) · 0ca594ed
  Neo authored Jun 12, 2025
  
  0ca594ed
- Support VILA models (#6106) · 31d6dee5
  Zijian authored Jun 12, 2025
  
  31d6dee5
- Fix misusing the "_is_cuda". (#7091) · 02543b54
  sogalin authored Jun 12, 2025
  
  02543b54
- Fix circular import in test_prefix_chunk_info.py (#7097) · 25a6a9aa
  Baizhou Zhang authored Jun 11, 2025
  
  25a6a9aa
- vlm: adapt internvl to VisionAttention (#6870) · 83d87685
  Mick authored Jun 11, 2025
  
  83d87685
- Fix GGuf and add back test_gguf.py (#7067) · 2a5f0100
  Baizhou Zhang authored Jun 10, 2025
  
  2a5f0100