Commits · 3c8ac78dc143a15eae5d6a4fdf44aec78ab27a80 · change / sglang · GitLab

03 Feb, 2025 1 commit
- optimize test_fused_moe style (#3268) · 3c8ac78d
  Xiaoyu Zhang authored Feb 03, 2025
  
  3c8ac78d
01 Feb, 2025 1 commit
- Add test for fp8 torch compile (#3246) · 53179026
  Ke Bao authored Feb 01, 2025
  
  53179026
31 Jan, 2025 1 commit
- [fix] Clamp logprob with dtype min to prevent `-inf` (#3224) · 734daedd
  Byron Hsu authored Jan 31, 2025
  
  734daedd
30 Jan, 2025 1 commit
- [test] Lower number of top logprobs to get rid of `-inf` (#3212) · 20453cef
  Byron Hsu authored Jan 30, 2025
  
  20453cef
28 Jan, 2025 3 commits
- [Fix] Address remaining issues of supporting MiniCPMV (#2977) · 9f635ea5
  Mick authored Jan 28, 2025
  
  9f635ea5
- [kernel] Use sgl_kernel rope (#3169) · 988d0a4b
  Byron Hsu authored Jan 27, 2025
```
Co-authored-by: zhyncs <me@zhyncs.com>
```
  988d0a4b
- [test] deduplicate test_session_control (#3183) · 27aeb4b7
  Byron Hsu authored Jan 27, 2025
  
  27aeb4b7
27 Jan, 2025 4 commits
- Update thresholds in test_nightly_gsm8k_eval.py (#3176) · f8ca66fb
  Lianmin Zheng authored Jan 27, 2025
  
  f8ca66fb
- Add activation parameters to fused_moe (#3170) · 52c03f16
  Lianmin Zheng authored Jan 27, 2025
  
  52c03f16
- add unit test for block wise fp8 (#3156) · 1e3e5215
  yizhang2077 authored Jan 27, 2025
  
  1e3e5215
- Add more logprob tests (#3162) · af02f99b
  Lianmin Zheng authored Jan 26, 2025
  
  af02f99b
26 Jan, 2025 3 commits
- Feature/function calling update (#2700) · b045841b
  YAMY authored Jan 26, 2025
```
Co-authored-by: Mingyuan Ma <mamingyuan2001@berkeley.edu>
Co-authored-by: Chayenne <zhaochen20@outlook.com>
Co-authored-by: shuaills <shishuaiuoe@gmail.com>
```
  b045841b
- Add a test case for cached_tokens (#3145) · d1a08632
  Lianmin Zheng authored Jan 26, 2025
  
  d1a08632
- Fix CI tests (#3132) · da6f8081
  Lianmin Zheng authored Jan 25, 2025
  
  da6f8081
21 Jan, 2025 1 commit
- Add accuracy and latency tests of eagle into CI (#3027) · a4331cd2
  Lianmin Zheng authored Jan 21, 2025
  
  a4331cd2
20 Jan, 2025 5 commits
- [Enhancement] Custom Logit Processor Improvement (#2998) · 583697cd
  Hongpeng Guo authored Jan 20, 2025
```
Signed-off-by: Hongpeng Guo <hpguo@anyscale.com>
```
  583697cd
- Skip flaky custom_logit_processor tests (#3004) · 51e87f6f
  Lianmin Zheng authored Jan 20, 2025
  
  51e87f6f
- Separate two entry points: Engine and HTTP server (#2996) · 03464890
  Lianmin Zheng authored Jan 19, 2025
```
Co-authored-by: fzyzcjy <5236035+fzyzcjy@users.noreply.github.com>
```
  03464890
- Improve metrics, logging, and importing orders (#2992) · cd493b5a
  Lianmin Zheng authored Jan 19, 2025
  
  cd493b5a
- Move sgl.Runtime under sglang/lang (#2990) · 61f42b57
  Lianmin Zheng authored Jan 19, 2025
  
  61f42b57
19 Jan, 2025 3 commits
- [Feature] Add sampler custom logits processor (#2396) · e403d237
  Hongpeng Guo authored Jan 19, 2025
```
Signed-off-by: Hongpeng Guo <hpguo@anyscale.com>
```
  e403d237
- support regex in xgrammar backend (#2983) · 3bcf5ece
  Enrique Shockwave authored Jan 19, 2025
  
  3bcf5ece
- Frontend: better error message handling for FINISH_ABORT in scheduler.py (#2956) · 4d4cdb3f
  Chang Su authored Jan 18, 2025
  
  4d4cdb3f
18 Jan, 2025 2 commits
- [Feature] Support minicpmv v2.6 (#2785) · 3d93f84a
  Mick authored Jan 19, 2025
```
Co-authored-by: Chayenne <zhaochen20@outlook.com>
Co-authored-by: yizhang2077 <1109276519@qq.com>
```
  3d93f84a
- support e4m3 kvcache in qwen2 & add kv scaling facotr json (#2894) · d3024f4f
  bjmsong authored Jan 18, 2025
```
Co-authored-by: bjmsong <bjmsong@126.com>
```
  d3024f4f
17 Jan, 2025 1 commit
- Add ut for qwen model (#2947) · d47c5101
  Ke Bao authored Jan 18, 2025
  
  d47c5101
16 Jan, 2025 2 commits
- [Frontend] Fix request length check and add option to disallow auto truncation in scheduler (#2876) · a8ccacc8
  Chang Su authored Jan 16, 2025
  
  a8ccacc8
- Improve benchmark scripts and error message printing (#2922) · 8f2c522a
  Lianmin Zheng authored Jan 16, 2025
  
  8f2c522a
15 Jan, 2025 2 commits
- adapt custom allreduce for tensorrt llm (#2511) · 767c9dec
  yizhang2077 authored Jan 16, 2025
  
  767c9dec
- Add ut for w8a8 int8 quantization (#2897) · bfbda62c
  Ke Bao authored Jan 15, 2025
  
  bfbda62c
13 Jan, 2025 7 commits
- CUDA-graph-compatible releasing and resuming KV cache and model weight memory (#2630) · 923f5183
  fzyzcjy authored Jan 14, 2025
  
  923f5183
- Revert "Integration of TurboMind AWQ" (#2866) · 6249e4a1
  Lianmin Zheng authored Jan 13, 2025
  
  6249e4a1
- Integration of TurboMind AWQ (#2828) · 17de02f9
  bjmsong authored Jan 13, 2025
```
Co-authored-by: root <bjmsong@126.com>
```
  17de02f9
- Collect more metrics: num_requests_total (#2859) · 51ab3ccf
  Lianmin Zheng authored Jan 13, 2025
  
  51ab3ccf
- Use only one GPU for MLA CI tests (#2858) · 67008f4b
  Lianmin Zheng authored Jan 13, 2025
  
  67008f4b
- Fix linear.py and improve weight loading (#2851) · 72c77763
  Lianmin Zheng authored Jan 13, 2025
```
Co-authored-by: SangBin Cho <rkooo567@gmail.com>
```
  72c77763
- Support FP8 E4M3 KV Cache (#2786) · 0bb0f763
  bjmsong authored Jan 13, 2025
```
Co-authored-by: root <bjmsong@126.com>
```
  0bb0f763
11 Jan, 2025 2 commits
- Improve: Token-In Token-Out Usage for RLHF (#2843) · c4f9707e
  Shi Shuai authored Jan 11, 2025
  
  c4f9707e
- Update threshold in test_nightly_gsm8k_eval.py (#2836) · f1769586
  Lianmin Zheng authored Jan 10, 2025
  
  f1769586
10 Jan, 2025 1 commit
- [Eagle2] Fix multiple concurrent request crashes (#2730) · a47bf391
  justdoit authored Jan 11, 2025
  
  a47bf391