Commits · a4c3b121d8aa92e84d3151694bed614a5d2ce85f · change / sglang · GitLab

29 Jul, 2025 14 commits
- Split the scheduler into multiple mixin classes to reduce the file size (#8483) · a4c3b121
  Lianmin Zheng authored Jul 29, 2025
  
  a4c3b121
- Fix moe align kernel test (#8531) · 5973675b
  Ke Bao authored Jul 30, 2025
  
  5973675b
- Update cutlass_moe.py (#8535) · 4d16c88b
  Elfie Guo authored Jul 29, 2025
  
  4d16c88b
- [sgl-kernel performace] fix fp8 quant kernels dispatch __nv_fp8_e4m3 bug to... · 7a4309cc
  Xiaoyu Zhang authored Jul 29, 2025
```
[sgl-kernel performace] fix fp8 quant kernels dispatch __nv_fp8_e4m3 bug to improve performance 10%-20% (#8499)
Co-authored-by: Ke Bao <ispobaoke@gmail.com>
```
  7a4309cc
- Update README.md (#8528) · 81367066
  Lianmin Zheng authored Jul 29, 2025
  
  81367066
- Always trigger pr-test (#8527) · 263c9236
  Lianmin Zheng authored Jul 29, 2025
  
  263c9236
- chore: bump v0.4.9.post6 (#8517) · 6478831b
  Yineng Zhang authored Jul 29, 2025
  
  6478831b
- Add PVC and update resource limits in k8s config (#8489) · 2e1d2d7e
  TimWang authored Jul 29, 2025
  
  2e1d2d7e
- Fix incorrect KV cache allocation for MTP models. (#8482) · fb16fbaf
  Lifu Huang authored Jul 28, 2025
```
Co-authored-by: Stefan He <hebiaobuaa@gmail.com>
```
  fb16fbaf
- Support colocating requests (#7973) · 0ce84c82
  fzyzcjy authored Jul 29, 2025
  
  0ce84c82
- Tiny add warnings for DeepEP when it is suboptimal (#8426) · 59d0bf01
  fzyzcjy authored Jul 29, 2025
  
  59d0bf01
- Reduce memory usage for fp4 moe (#8413) · 7df2c0c2
  fzyzcjy authored Jul 29, 2025
  
  7df2c0c2
- Rename the last step in pr-test.yml as pr-test-finish (#8486) · 69712e6f
  Lianmin Zheng authored Jul 28, 2025
  
  69712e6f
- Update CODEOWNERS (#8485) · 001bffca
  Lianmin Zheng authored Jul 28, 2025
  
  001bffca
28 Jul, 2025 26 commits
- [CI]Add genai-bench Performance Validation for PD Router (#8477) · 7c969717
  Keyang Ru authored Jul 28, 2025
```
Co-authored-by: key4ng <rukeyang@gamil.com>
```
  7c969717
- 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
- Update codeowner (#8476) · bd516949
  Lianmin Zheng authored Jul 28, 2025
  
  bd516949
- 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
- feat: support glm4 tuning (#8473) · 1466c1b8
  Yineng Zhang authored Jul 28, 2025
  
  1466c1b8
- [3/N] MoE Refactor: Simplify DeepEP Output (#8421) · 9c138a04
  Cheng Wan authored Jul 28, 2025
  
  9c138a04
- Fix parsing ChatCompletionMessage (#7273) · c8f549d9
  Timofey authored Jul 28, 2025
```
Co-authored-by: Timofey K <timosha1113@gmail.com>
```
  c8f549d9
- [NVIDIA] Change to use `num_local_experts` (#8453) · 134fa43e
  Kaixi Hou authored Jul 28, 2025
  
  134fa43e
- fix: update dep (#8467) · ccfe52a0
  Yineng Zhang authored Jul 28, 2025
  
  ccfe52a0
- feat: throttle requests at scheduler based on --max_queued_requests (#7565) · 747dd450
  harrisonlimh authored Jul 28, 2025
  
  747dd450
- Update PR template (#8465) · b5821592
  Ke Bao authored Jul 28, 2025
  
  b5821592
- fix:reorder topk experts to ensure shared expert replaces minimal score (#8125) · a9dd3ec3
  erictanjn authored Jul 28, 2025
  
  a9dd3ec3
- chore: bump v0.4.9.post5 (#8458) · 45bc170b
  Yineng Zhang authored Jul 28, 2025
  
  45bc170b
- Revert "[kernel] opt moe align block kernel by block/warp scan algorithm" (#8457) · 22623699
  Xiaoyu Zhang authored Jul 28, 2025
  
  22623699
- Fix per_token_group_quant_8bit when hidden_dim // group_size is not divided by 4. (#8449) · fb4ce17d
  strgrb authored Jul 28, 2025
```
Co-authored-by: Zhang Kaihong <zhangkaihong.zkh@alibaba-inc.com>
```
  fb4ce17d
- fix GLM4_MOE launch with compressed_tensor quant model (#8456) · 25f73c6c
  Minglei Zhu authored Jul 28, 2025
  
  25f73c6c
- GLM-4.5 Model Support Follow-up (#8445) · 581e7dcb
  Binyao Jiang authored Jul 27, 2025
  
  581e7dcb
- doc: add bench_one_batch_server in the benchmark doc (#8441) · 484d0e02
  Qiaolin Yu authored Jul 27, 2025
  
  484d0e02
- Remove zstd compression for building Dockerfile.gb200 (#8442) · 5922c0cb
  kyleliang-nv authored Jul 27, 2025
  
  5922c0cb
- GLM-4.5 Model Support (#8224) · 6d6a8bc2
  Yuxuan Zhang authored Jul 28, 2025
```
Co-authored-by: Lifu Huang <lifu.hlf@gmail.com>
Co-authored-by: Binyao Jiang <byjiang1996@gmail.com>
Co-authored-by: Stefan He <hebiaobuaa@gmail.com>
```
  6d6a8bc2
- [PD] Fix abort_request for PD disaggregation (#8352) · 2fd5c704
  Shangming Cai authored Jul 28, 2025
```
Signed-off-by: Shangming Cai <caishangming@linux.alibaba.com>
Co-authored-by: ybyang <10629930+whybeyoung@users.noreply.github.com>
```
  2fd5c704
- chore: bump transformer to 4.54.0 (#8416) · 4ad97370
  Stefan He authored Jul 27, 2025
```
Co-authored-by: Binyao Jiang <byjiang1996@gmail.com>
Co-authored-by: Lifu Huang <lifu.hlf@gmail.com>
```
  4ad97370
- [feat] Support different attention backends for prefill and decode (#6338) · 28103384
  Qiaolin Yu authored Jul 27, 2025
```
Co-authored-by: tianqilin.99 <tianqilin.99@bytedance.com>
Co-authored-by: Baizhou Zhang <sobereddiezhang@gmail.com>
```
  28103384
- [router] improve router logs and request id header (#8415) · fe6a445d
  Simo Lin authored Jul 27, 2025
  
  fe6a445d
- bugfix: Fix XGrammar backend to use model's EOS tokens for constrained generation (#8422) · dd487e55
  Chang Su authored Jul 27, 2025
  
  dd487e55