Commits · a5375adc3a8f87040a632f0b5aeb19c336ff87fc · change / sglang

"src/vscode:/vscode.git/clone" did not exist on "2238386ad07c10fd13d3fd64c65857c8cd23b5cd"

17 Feb, 2025 6 commits
- chore: bump v0.4.3.post2 (#3645) · a5375adc
  Yineng Zhang authored Feb 18, 2025
```
Co-authored-by: pankajroark <pankajroark@users.noreply.github.com>
```
  a5375adc
- chore: update flashinfer v0.2.1.post2 (#3644) · 75d171a9
  Yineng Zhang authored Feb 18, 2025
  
  75d171a9
- feat: support flashinfer mla with prefix cache (#3643) · 714f3e63
  Yineng Zhang authored Feb 18, 2025
  
  714f3e63
- support multi-gpu block-gemm tuning (#3639) · c38f3aed
  Xiaoyu Zhang authored Feb 18, 2025
  
  c38f3aed
- chore: bump v0.4.3.post1 (#3638) · e782eb7e
  Yineng Zhang authored Feb 17, 2025
  
  e782eb7e
- Revert "[ROCm] Use `tl.range()` in block GEMM kernels with `num_stage… (#3632) · 5f1a485d
  Yineng Zhang authored Feb 17, 2025
  
  5f1a485d
16 Feb, 2025 3 commits
- [ROCm] Use `tl.range()` in block GEMM kernels with `num_stages` set by host. (#3535) · 03caefeb
  Wen-Heng (Jack) Chung authored Feb 16, 2025
```
Co-authored-by: HAI <hixiao@gmail.com>
```
  03caefeb
- Model: Support Qwen 2.5 vl (#3258) · bcc213df
  Mick authored Feb 16, 2025
  
  bcc213df
- fix lockfile and port_registry file permission error (#3598) · 39416e39
  Jiada Li authored Feb 15, 2025
```
Co-authored-by: jiada li <jiada@lmsys.us-northcentral1-a.compute.internal>
Co-authored-by: zhaochenyang20 <zhaochen20@outlook.com>
```
  39416e39
15 Feb, 2025 4 commits
- fix apply_token_bitmask_inplace_cuda (#3594) · bbc47c34
  Yineng Zhang authored Feb 15, 2025
  
  bbc47c34
- fix high qps crash when enable mtp (#3592) · dfce9269
  Yineng Zhang authored Feb 15, 2025
```
Co-authored-by: ispobock <ispobaoke@hotmail.com>
```
  dfce9269
- doc: emphasize and notify the usage of chat_template (#3589) · 7711ac6e
  Mick authored Feb 15, 2025
```
Co-authored-by: Chayenne <zhaochen20@outlook.com>
```
  7711ac6e
- [CI] Improve Docs CI Efficiency (#3587) · 7443197a
  Shi Shuai authored Feb 15, 2025
```
Co-authored-by: zhaochenyang20 <zhaochen20@outlook.com>
```
  7443197a
14 Feb, 2025 5 commits
- Support NextN (MTP) speculative decoding for DeepSeek-V3/R1 (#3582) · 862dd76c
  Ke Bao authored Feb 15, 2025
  
  862dd76c
- [fix] added support for vlm in offline inference (#3548) · fb4c9c3a
  Shenggui Li authored Feb 15, 2025
  
  fb4c9c3a
- Add support for OpenAI API o1 model (#3363) · 6cc30955
  Chuyue Sun authored Feb 13, 2025
```
Co-authored-by: Shan Yu <shanyu1@g.ucla.edu>
```
  6cc30955
- chore: bump v0.4.3 (#3556) · e0b9a423
  Yineng Zhang authored Feb 14, 2025
  
  e0b9a423
- feat: support flashinfer mla attention for deepseek v3 (#3550) · 70f894b8
  Yineng Zhang authored Feb 14, 2025
  
  70f894b8
13 Feb, 2025 1 commit
- [ROCm] Add ROCm tuning configs for AMD Instinct MI325X. (#3536) · 871a4aa1
  Wen-Heng (Jack) Chung authored Feb 12, 2025
  
  871a4aa1
12 Feb, 2025 3 commits
- integrate blockwise fp8 kernel (#3529) · 98eecbda
  yizhang2077 authored Feb 13, 2025
  
  98eecbda
- Fix deepseek awq v3 (#3450) · 8616357a
  Liangsheng Yin authored Feb 12, 2025
  
  8616357a
- use sgl_per_token_group_quant_fp8 kernel (#3493) · 45e3a7bc
  Xiaoyu Zhang authored Feb 12, 2025
  
  45e3a7bc
11 Feb, 2025 5 commits
- Make NCCL NVLS configurable (#3502) · b8318aec
  Ata Fatahi authored Feb 11, 2025
  
  b8318aec
- MI30x: More graph captures for larger batch sizes and concurrencies (#3420) · d81ac443
  HAI authored Feb 11, 2025
  
  d81ac443
- Support Eagle cuda graph for Triton backend (#3500) · 7e6d5fc6
  Ke Bao authored Feb 12, 2025
  
  7e6d5fc6
- Tune MI300X fused MoE Triton kernel JSON config. (#3492) · cadd5dbe
  Wen-Heng (Jack) Chung authored Feb 11, 2025
  
  cadd5dbe
- [ROCm] Add ROCm tuning config to block gemm and Re-tune for AMD Radeon Graphics (#3418) · fdf04a14
  yigex authored Feb 11, 2025
```
Co-authored-by: Bruce Xue <yigex@xilinx.com>
Co-authored-by: HAI <hixiao@gmail.com>
```
  fdf04a14
10 Feb, 2025 7 commits
- [Feat] Return hidden states (experimental) (#3364) · 5f0e7de3
  Jackmin801 authored Feb 10, 2025
```
Co-authored-by: Chayenne <zhaochen20@outlook.com>
```
  5f0e7de3
- refine some typo (#3473) · 2f47d710
  Xiaoyu Zhang authored Feb 10, 2025
  
  2f47d710
- [Eagle] reduce one draft forward (#3468) · d23cb9a0
  Ying Sheng authored Feb 10, 2025
  
  d23cb9a0
- Support Eagle2 for Triton backend (#3466) · 2d611323
  Ke Bao authored Feb 10, 2025
  
  2d611323
- chore: bump v0.4.2.post4 (#3459) · cddb1cdf
  Yineng Zhang authored Feb 10, 2025
  
  cddb1cdf
- [Fix] Fix accuracy bug and refactor codes for lora (#3413) · c45cab1c
  Baizhou Zhang authored Feb 09, 2025
  
  c45cab1c
- remove _grouped_size_compiled_for_decode_kernels (#3453) · 27c4c9cf
  Yineng Zhang authored Feb 10, 2025
  
  27c4c9cf
09 Feb, 2025 6 commits
- feat: enable ragged fa3 by default on hopper 12.4+ (#3442) · 36f6fc50
  Yineng Zhang authored Feb 10, 2025
  
  36f6fc50
- support version in sgl-kernel (#3439) · 4cfd3add
  Yineng Zhang authored Feb 10, 2025
  
  4cfd3add
- compatible with new outlines (#3435) · 85986bb9
  Yineng Zhang authored Feb 10, 2025
  
  85986bb9
- remove activation dependency in fused_moe (#3433) · 64c87135
  Yineng Zhang authored Feb 10, 2025
  
  64c87135
- fix draft cuda graph capture failure (#3431) · 1646149a
  Yineng Zhang authored Feb 09, 2025
  
  1646149a
- add cuda graph capture failure possible solution (#3430) · bc72e5bd
  Yineng Zhang authored Feb 09, 2025
  
  bc72e5bd