Commits · 7a1f7fc5049d4fd5b3adc88fe48466fccbe676d8 · zhaoyu6 / sglang · GitLab

31 Jul, 2025 3 commits
- [Feature] Hybrid EP and TP (#8590) · 7a1f7fc5
  Cheng Wan authored Jul 31, 2025
  
  7a1f7fc5
- [4/N] MoE Refactor: Unified Triton Kernel for FusedMoE and EPMoE (#8515) · 32fa1e9c
  Cheng Wan authored Jul 31, 2025
  
  32fa1e9c
- update sgl-kernel for EP: python part (#8550) · e179e0b7
  Cheng Wan authored Jul 31, 2025
  
  e179e0b7
30 Jul, 2025 1 commit
- Update cutlass_moe.py (#8545) · e3f08c77
  Elfie Guo authored Jul 29, 2025
  
  e3f08c77
29 Jul, 2025 3 commits
- Support EPLB in FusedMoE (#8448) · 9effeb5b
  Cheng Wan authored Jul 29, 2025
  
  9effeb5b
- Update cutlass_moe.py (#8535) · 4d16c88b
  Elfie Guo authored Jul 29, 2025
  
  4d16c88b
- Tiny add warnings for DeepEP when it is suboptimal (#8426) · 59d0bf01
  fzyzcjy authored Jul 29, 2025
  
  59d0bf01
28 Jul, 2025 6 commits
- chore: add glm 4.5 fp8 tp4 config (#8480) · 8240a6b0
  Yineng Zhang authored Jul 28, 2025
  
  8240a6b0
- chore: add glm4 fp8 tp8 config (#8478) · 3a04aa4b
  Yineng Zhang authored Jul 28, 2025
  
  3a04aa4b
- Fix DEEPEP BF16 compatibility for Deepseek Style model like GLM 4.5 (#8469) · 74e7e457
  Stefan He authored Jul 28, 2025
```
Co-authored-by: Minglei Zhu <mingleizhu1122@gmail.com>
```
  74e7e457
- [3/N] MoE Refactor: Simplify DeepEP Output (#8421) · 9c138a04
  Cheng Wan authored Jul 28, 2025
  
  9c138a04
- [NVIDIA] Change to use `num_local_experts` (#8453) · 134fa43e
  Kaixi Hou authored Jul 28, 2025
  
  134fa43e
- fix:reorder topk experts to ensure shared expert replaces minimal score (#8125) · a9dd3ec3
  erictanjn authored Jul 28, 2025
  
  a9dd3ec3
27 Jul, 2025 5 commits
- Support triton kernels v3.4.0 for fused_moe (#8258) · b3eac168
  Yuan Luo authored Jul 27, 2025
```
Co-authored-by: luoyuan.luo <luoyuan.luo@antgroup.com>
Co-authored-by: Cheng Wan <cwan@x.ai>
Co-authored-by: Cheng Wan <54331508+ch-wan@users.noreply.github.com>
```
  b3eac168
- [hotfix] fix merge conflicts in FlashInferEPMoE (#8405) · 4d921f2b
  Cheng Wan authored Jul 27, 2025
  
  4d921f2b
- [2/N] MoE Refactor: Unify weight loader and quant methods (#8397) · bf0f448f
  Cheng Wan authored Jul 27, 2025
  
  bf0f448f
- Add A800 fused MoE kernel tuning configs for Qwen3-Coder-480B-A35B-Instruct (#8351) · 2a1936de
  Li Hui authored Jul 27, 2025
  
  2a1936de
- [NVIDIA] Add Flashinfer MoE blockscale fp8 backend (#8036) · 85486b6f
  Kaixi Hou authored Jul 27, 2025
  
  85486b6f
25 Jul, 2025 3 commits
- Clean up server_args, triton cache manager (#8332) · ed2e313e
  Lianmin Zheng authored Jul 25, 2025
  
  ed2e313e
- Add H20-3e fused MoE kernel tuning configs for Qwen3-Coder-480B-A35B-Instruct (#8344) · 12cb760a
  Xu Wenqing authored Jul 26, 2025
```
Signed-off-by: 许文卿 <xwq391974@alibaba-inc.com>
```
  12cb760a
- [torch.compile bug] avoid biased_grouped_topk_impl func repeatedly triggering... · 9045cc1e
  Xiaoyu Zhang authored Jul 25, 2025
```
[torch.compile bug] avoid biased_grouped_topk_impl func repeatedly triggering `torch.compile` in forward pass (#8353)
```
  9045cc1e
24 Jul, 2025 1 commit
- [code style] Clean dead triton kernel code in fused_moe and useless vllm_ops import (#8310) · a167fd0b
  Xiaoyu Zhang authored Jul 24, 2025
  
  a167fd0b
22 Jul, 2025 1 commit
- [AMD] Remove vllm's scaled_fp8_quant and moe_sum when SGLANG_USE_AITER=1 (#7484) · e50109f2
  Hubert Lu authored Jul 21, 2025
  
  e50109f2
21 Jul, 2025 1 commit
- Apply fused sorted token ids padding (#8193) · c9e8613c
  Ke Bao authored Jul 21, 2025
  
  c9e8613c
20 Jul, 2025 5 commits
- Fix dtype error in CI (#8197) · 465968b2
  Ke Bao authored Jul 21, 2025
  
  465968b2
- feat: add h200 tp 16 kimi k2 moe config (#8183) · bbcfbc1a
  Qiaolin Yu authored Jul 20, 2025
  
  bbcfbc1a
- fix moe gate dtype, fix tbo, fix fake dispatch (#7825) · a589a071
  Atream authored Jul 20, 2025
  
  a589a071
- feat: add b200 tp 16 kimi k2 moe config (#8178) · f62d75b6
  Yineng Zhang authored Jul 19, 2025
  
  f62d75b6
- feat: add h200 tp 16 kimi k2 moe config (#8176) · 0f9b11e3
  Yineng Zhang authored Jul 19, 2025
  
  0f9b11e3
19 Jul, 2025 1 commit
- [1/N] MoE Refactor: refactor `select_experts` (#7966) · 15ad6c90
  Cheng Wan authored Jul 19, 2025
  
  15ad6c90
18 Jul, 2025 1 commit
- [CPU][Llama4] Fix Llama4 MoE inputs with "apply_router_weight_on_input" (#7889) · 48c1fa7b
  jianan-gu authored Jul 18, 2025
  
  48c1fa7b
17 Jul, 2025 2 commits
- Refactor: move all quantization-related code to `srt/layer/quantization` (#7989) · 49b87774
  Cheng Wan authored Jul 17, 2025
  
  49b87774
- [Fix] ensure DeepGEMM is only enabled for FP8_W8A8 models (#8110) · 5c08a36c
  hzh0425 authored Jul 17, 2025
  
  5c08a36c
16 Jul, 2025 1 commit
- [1/n] chore: decouple quantization implementation from vLLM dependency (#7992) · c28ad199
  Peng Zhang authored Jul 17, 2025
  
  c28ad199
15 Jul, 2025 1 commit
- H20 tune config for Kimi (#8047) · 14f1f151
  yhang authored Jul 16, 2025
  
  14f1f151
14 Jul, 2025 1 commit
- perf: add kimi k2 fused_moe tuning config for h30_3e (#8021) · c07f647c
  Yusong Gao authored Jul 14, 2025
```
Co-authored-by: yudian0504 <yudian.zy@antgroup.com>
```
  c07f647c
12 Jul, 2025 1 commit
- [BugFix] fix pre_reorder_triton_kernel default int32 issue (#7814) · 5f6756b0
  Morpheus Guo authored Jul 13, 2025
  
  5f6756b0
11 Jul, 2025 1 commit
- fix: minor fix for modelopt weight load compatibility (#7953) · 191d836f
  Peng Zhang authored Jul 12, 2025
  
  191d836f
10 Jul, 2025 1 commit

[feature]Ascend quantization support (#7791) · 766392c6

ronnie_zheng authored Jul 10, 2025


Co-authored-by: ichernob <ichernobnn@gmail.com>
Co-authored-by: liupeng <liupeng374@huawei.com>

766392c6

09 Jul, 2025 1 commit
- [CPU][Qwen3 MoE] Enable fused_topk CPU fusion and enhance FP8 TP padding (#7838) · d389bedf
  jianan-gu authored Jul 09, 2025
  
  d389bedf