Commits · 4bd2952a376a7781526ffb063536b90e38b09573 · zhaoyu6 / sglang

"tests/vscode:/vscode.git/clone" did not exist on "5db9abde439bc02c3791da2a4fefee80d94d5b96"

16 May, 2025 4 commits
- feat: add dp attention support for Qwen 2/3 MoE models, fixes #6088 (#6121) · 4bd2952a
  Fr4nk1in authored May 17, 2025
```
Co-authored-by: King.Zevin <zevin@mail.ustc.edu.cn>
Co-authored-by: Yi Zhang <1109276519@qq.com>
```
  4bd2952a
- [2/2] Add python wrapper for CUTLASS FP8 Blockscale MoE Kernel. (#5694) · 6fc93575
  Elfie Guo authored May 16, 2025
  
  6fc93575
- [Fix] Improve dependencies for Blackwell image (#6334) · 839fb31e
  Baizhou Zhang authored May 16, 2025
  
  839fb31e
- Support precomputed multimodal features for Qwen-VL and Gemma3 models. (#6136) · f19a9204
  Yury Sulsky authored May 16, 2025
```
Co-authored-by: Yury Sulsky <ysulsky@tesla.com>
```
  f19a9204
15 May, 2025 8 commits
- Minor improvements of TokenizerManager / health check (#6327) · e07a6977
  Lianmin Zheng authored May 15, 2025
  
  e07a6977
- Fix lora bench (#6302) · cd8d4b9d
  Qiaolin Yu authored May 15, 2025
  
  cd8d4b9d
- Reduce MoE memory usage (#6147) · f194e14f
  fzyzcjy authored May 16, 2025
  
  f194e14f
- Fix gpu mem check on CPU (#6317) · cfc9f9ab
  Yi Liu authored May 16, 2025
```
Signed-off-by: yiliu30 <yi4.liu@intel.com>
```
  cfc9f9ab
- [misc] remove redundant platform codes (#6298) · 9a405274
  JieXin Liang authored May 15, 2025
  
  9a405274
- [Feat] Support FlashMLA backend with MTP and FP8 KV cache (#6109) · 2e4babdb
  quinnrong94 authored May 15, 2025
```
Co-authored-by: Yingyi <yingyihuang2000@outlook.com>
Co-authored-by: neiltian <neiltian@tencent.com>
Co-authored-by: lukec <118525388+sleepcoo@users.noreply.github.com>
Co-authored-by: kexueyu <kexueyu@tencent.com>
Co-authored-by: vincentmeng <vincentmeng@tencent.com>
Co-authored-by: pengmeng <pengmeng@tencent.com>
```
  2e4babdb
- [fix][RL] Remove the incorrect barrier in init_weights_update_group (#5914) · 44a3783d
  Zilin Zhu authored May 15, 2025
  
  44a3783d
- feat: add flush cache to EngineBase and HttpServerEngineAdapter (#6009) · f3bf6110
  Junrong Lin authored May 15, 2025
  
  f3bf6110
14 May, 2025 4 commits
- [AMD] Fix Llama 4 Scout and Maverick accuracy issues on MI300X (#6274) · 198b9056
  Hubert Lu authored May 14, 2025
  
  198b9056
- [Bug] Fix accidental logger override caused by internVL. (#6282) · 3e350a93
  Lifu Huang authored May 13, 2025
  
  3e350a93
- Fix a bug in schedule_policy (#6276) · fb71725c
  Ying Sheng authored May 13, 2025
  
  fb71725c
- perf: optimize local_block_table memory allocation (#6273) · 912788c0
  Chang Su authored May 13, 2025
  
  912788c0
13 May, 2025 6 commits
- chore: bump v0.4.6.post4 (#6245) · 16267d4f
  Yineng Zhang authored May 13, 2025
  
  16267d4f
- [misc] deep_gemm fallback to NVRTC when NVCC not found (#6252) · 17299f08
  JieXin Liang authored May 13, 2025
  
  17299f08
- model(vlm): pixtral (#5084) · 5380cd7e
  Kiv Chen authored May 13, 2025
  
  5380cd7e
- Fix two issues related to `--moe-dense-tp-size=1` (#5657) · b2e95f62
  Cheng Wan authored May 13, 2025
```
Co-authored-by: liusy58 <liusy58@linux.alibaba.com>
Co-authored-by: 颉沆 <xiehang.lsy@alibaba-inc.com>
```
  b2e95f62
- [VERL Use Case] Add torch_memory_saver into deps (#6247) · 1ab14c4c
  Stefan He authored May 12, 2025
  
  1ab14c4c
- Skip the flaky test_stateful_custom_logit_processor (#6251) · ac2324c1
  Lianmin Zheng authored May 12, 2025
  
  ac2324c1
12 May, 2025 11 commits
- fix typo (#6248) · f24fc5b8
  Yineng Zhang authored May 12, 2025
  
  f24fc5b8
- Support incremental streaming of logprob/token_ids between scheduler and detokenizer (#6225) · d18c6b33
  Lianmin Zheng authored May 12, 2025
```
Co-authored-by: SangBin Cho <rkooo567@gmail.com>
```
  d18c6b33
- [PD] Add support for different TP sizes per DP rank (#5922) · f1c89600
  shangmingc authored May 13, 2025
```
Signed-off-by: Shangming Cai <caishangming@linux.alibaba.com>
```
  f1c89600
- chore: add hf_xet dep (#6243) · f94543d2
  Yineng Zhang authored May 12, 2025
  
  f94543d2
- Revert "fix some typos" (#6244) · e8e18dcd
  Lianmin Zheng authored May 12, 2025
  
  e8e18dcd
- [PP] Fix init_memory_pool desync & add PP for mixtral (#6223) · bad7c26f
  Ying Sheng authored May 12, 2025
  
  bad7c26f
- fix some typos (#6209) · d738ab52
  applesaucethebun authored May 12, 2025
```
Co-authored-by: Brayden Zhong <b8zhong@uwaterloo.ca>
```
  d738ab52
- Log if cuda graph is used & extend cuda graph capture to cuda-graph-max-bs (#6201) · fba8eccd
  Lianmin Zheng authored May 12, 2025
```
Co-authored-by: SangBin Cho <rkooo567@gmail.com>
```
  fba8eccd
- Performing Vocabulary Parallelism for LM Head across Attention TP Groups (#5558) · 25c83fff
  Cheng Wan authored May 12, 2025
```
Co-authored-by: liusy58 <liusy58@linux.alibaba.com>
```
  25c83fff
- [doc] add a note for --n-share-experts-fusion args (#6154) · 9f2c9568
  Xiaoyu Zhang authored May 12, 2025
  
  9f2c9568
- Fix start_profile does not support with_stack and record_shapes (#6043) · 3f2702ae
  fzyzcjy authored May 12, 2025
  
  3f2702ae
11 May, 2025 7 commits
- Replace time.time() to time.perf_counter() for benchmarking. (#6178) · 6e2da515
  Lifu Huang authored May 11, 2025
```
Signed-off-by: Lifu Huang <lifu.hlf@gmail.com>
```
  6e2da515
- [CI] Reorganize the 8 gpu tests (#6192) · 03227c5f
  Lianmin Zheng authored May 11, 2025
  
  03227c5f
- Improve structured outputs: fix race condition, server crash, metrics and style (#6188) · 01bdbf7f
  Lianmin Zheng authored May 11, 2025
  
  01bdbf7f
- Handle empty input string for embedding models (#5621) · 41a645f5
  Ravi Theja authored May 11, 2025
```
Co-authored-by: Ravi Theja Desetty <ravitheja@Ravis-MacBook-Pro.local>
```
  41a645f5
- chore: upgrade sgl-kernel v0.1.2.post1 (#6196) · 23010630
  Yineng Zhang authored May 11, 2025
```
Co-authored-by: alcanderian <alcanderian@gmail.com>
```
  23010630
- fix: handle None multimodal_inputs during merging and filtering batches in... · 41273fd7
  Yusong Gao authored May 11, 2025
```
fix: handle None multimodal_inputs during merging and filtering batches in disaggregation decode mode (#6169)
```
  41273fd7
- [fix] remove mixtral from is_fa3_default_architecture (#6191) · e9bebafb
  JieXin Liang authored May 11, 2025
  
  e9bebafb