Commit Graph

2461 Commits

Author SHA1 Message Date
Stefan He
ef3c2dd08e Support Online Quantization for W8A8 (#4485) 2025-03-17 00:28:56 -07:00
Wenbo Yang
75b656488a Support serving DeepSeek-R1-Channel-INT8 with 32 L40S. (#4418) 2025-03-17 00:03:43 -07:00
Mick
0f52fb55ec config: Update fused moe config (#4493) 2025-03-16 23:51:58 -07:00
萝卜菜
d6d21640d3 [Feature] Support Deepseek-VL2 (#2798)
Co-authored-by: Edenzzzz <wtan45@wisc.edu>
Co-authored-by: Chayenne <zhaochen20@outlook.com>
Co-authored-by: Yi Zhang <1109276519@qq.com>
2025-03-16 23:07:59 -07:00
JieXin Liang
0212d2e288 [Fix] use torch.inference_mode() instead of torch.no_grad() (#4372) 2025-03-16 22:54:16 -07:00
Byron Hsu
8cc300f536 Fix router test (#4483) 2025-03-16 22:49:47 -07:00
mlmz
452db50808 Constraint Decoding: Set xgrammar as the default grammar backend (#4386) 2025-03-16 18:53:43 -07:00
Rin Intachuen
d1112d8548 Add endpoint for file support, purely to speed up processing of input_embeds. (#2797) 2025-03-16 18:30:37 -07:00
woodx
48efec7b05 Feature: support code completion (#3612) 2025-03-16 18:26:19 -07:00
yiakwy-xpu-ml-framework-team
9b8333d992 [ROCm] enable moe topk softmax in amd (#4448) 2025-03-16 18:16:55 -07:00
Zhiqiang Xie
f5bbf6037d Fix: Complete int32 to int64 conversion (#4465) 2025-03-16 18:14:27 -07:00
huiwq1990
5cbd709ea1 Fix: modelscope env comment (#4474)
Signed-off-by: huiwq1990 <huiwq1990@163.com>
2025-03-16 18:11:33 -07:00
Yinghai Lu
2e4a1e2d05 Initialize image processor for skip-tokenizer-init codepath (#4479)
Co-authored-by: Alex Kirillov <alex@iterationlab.org>
2025-03-16 18:10:09 -07:00
Mick
9d02bb3e2a Urgent model support: support gemma-3-it (#4424) 2025-03-16 17:37:32 -07:00
ZelinTan
402db5c58c Benchmark: Statistical Analysis of the Output Stability of the Deepseek Model (#4202)
Co-authored-by: Chayenne <zhaochen20@outlook.com>
2025-03-16 17:32:57 -07:00
Lianmin Zheng
754a0e8278 Update CODEOWNERS (#4484) 2025-03-16 17:10:15 -07:00
Yinghai Lu
799fb5f455 400 on empty input_ids (#4481) 2025-03-16 14:01:23 -07:00
Yi Zhang
25e1816eff fix custom allreduce performance/accuracy problem (#4477) 2025-03-16 12:16:30 -07:00
lukec
a53fe428f9 Support FlashMLA backend (#4472)
Co-authored-by: yinfan98 <1106310035@qq.com>
2025-03-16 09:07:06 -07:00
Ying Sheng
1b859295f4 [Eagle] Remove the greedy branch and some redundant code (#4363)
Co-authored-by: Sehoon Kim <sehoon@x.ai>
2025-03-16 02:48:55 -07:00
Yineng Zhang
9971dc2283 Revert "feat: Add FlashMLA submodule (#4449)" (#4470) 2025-03-16 01:30:05 -07:00
Lianmin Zheng
3db35c1af4 Release sgl-kernel v0.0.5.post2 (#4469) 2025-03-16 01:01:53 -07:00
Ying Sheng
52a34d7448 Add greedy verification kernel (#4383) 2025-03-16 00:58:26 -07:00
Lianmin Zheng
06d12b39d3 Remove filter for pr-tests (#4468) 2025-03-16 00:57:26 -07:00
Lianmin Zheng
c30976fb41 Fix finish step for pr tests and notebook tests (#4467) 2025-03-16 00:52:06 -07:00
JieXin Liang
1a3fa75f2f [Fix] use torch.cat instead of torch.concat to prevent entering the Autograd backends. (#4466) 2025-03-16 00:02:47 -07:00
Shi Shuai
81f431eded feat: Add FlashMLA submodule (#4449)
Co-authored-by: sleepcoo <sleepcoo@gmail.com>
2025-03-15 23:30:25 -07:00
Yineng Zhang
65b7c9b78f cleanup deps 2/n (#4464) 2025-03-15 23:06:17 -07:00
Lianmin Zheng
2c4f5ccac1 Fix minor style (#4460) 2025-03-15 21:51:12 -07:00
Wang Ran (汪然)
158430473e Fix typos (#4368) 2025-03-15 21:27:58 -07:00
Mick
8ec2ce0726 perf: update fused moe config (#4459) 2025-03-15 21:23:57 -07:00
Michael Feil
1fd0cf8a7b Update comment in qwen2.py (#4447) 2025-03-15 21:14:29 -07:00
vikram singh shekhawat
bf63ee54ed Auto-detect device if not specified in server arguments. (#4423) 2025-03-15 21:13:51 -07:00
Wang Ran (汪然)
22c96f78a6 typos: Update sampling_params.md (#4391)
Co-authored-by: Lianmin Zheng <lianminzheng@gmail.com>
2025-03-15 16:40:18 -07:00
Wang Ran (汪然)
2892b9bb97 bugfix: Update sampling_params.py (#4413) 2025-03-15 16:39:19 -07:00
Xu Song
470b474075 Update bench_serving.py (#4454) 2025-03-15 16:33:58 -07:00
江家瑋
26c372c13c docs: Add Llama 3.3 to supported models (#4453)
Signed-off-by: JiangJiaWei1103 <waynechuang97@gmail.com>
2025-03-15 16:33:43 -07:00
Chen Shengzhi
86d9baedc2 [Fix] Fix errors when using the device except cuda. (#4455) 2025-03-15 16:33:00 -07:00
lukec
21d485f835 Fix test_create_kvindices unit test (#4452) 2025-03-15 16:01:04 -07:00
Mick
035ac2ab74 ci: update transformers==4.48.3 (#4451) 2025-03-15 13:27:26 -07:00
Chayenne
e1a5e7e47d docs: hot fix torch compile cache (#4442) 2025-03-14 19:05:59 -07:00
Yineng Zhang
ad1ae7f7cd use topk_softmax with sgl-kernel (#4439) 2025-03-14 15:59:06 -07:00
Lianmin Zheng
e73167ade3 Fix maximum recursion depth triggered on exception exit (#4438) 2025-03-14 15:12:26 -07:00
Yineng Zhang
862fe52241 bump v0.0.5.post1 (#4437) 2025-03-14 15:00:26 -07:00
Qingquan Song
61e4433caf Add moe topk softmax templated from vllm (#4302) 2025-03-14 12:03:33 -07:00
Zhan Lu
660305c38a [Doc] fix wrong flag in deepseek documentation (#4427) 2025-03-14 11:30:55 -07:00
Baoyuan Qi
642ab418f3 [bug] fix duplicate variable MAX_PIXELS in qwen_vl.py (#4419) 2025-03-14 01:28:25 -07:00
wangyu
1ce4878d31 feat(remote_model): support variable remote backend for model loader (#3964)
Signed-off-by: wangyu <wangyu.steph@bytedance.com>
2025-03-14 00:40:44 -07:00
Yineng Zhang
977d7cd26a cleanup deps 1/n (#4400)
Co-authored-by: sleepcoo <sleepcoo@gmail.com>
2025-03-14 00:00:33 -07:00
Lu Changqi
0e0ec70200 Hierarchical Caching supports MLA (#4009)
Signed-off-by: Changqi Lu <luchangqi.123@bytedance.com>
Co-authored-by: Zhiqiang Xie <xiezhq@stanford.edu>
2025-03-13 20:42:14 -07:00