anodyine/vllm-omni
1
0 Fork 0
Code Issues 4 Pull Requests 2 Actions Packages Projects Releases Wiki Activity
453 Commits 235 Branches 31 Tags
b4775d52fed7692ea464c8fbcd5689b16977c840
Commit Graph
453 Commits
This Branch
This Branch
All Branches
Author SHA1 Message Date
copilot-swe-agent[bot]andhsliuustc0106 b4775d52fe Extract usage info from text output before modality filtering 
This ensures that usage/billing data is captured even when text output
is filtered out (e.g., for audio-only requests). Previously, audio-only
requests would return zero usage tokens because the text output was
skipped before extracting usage information.

Co-authored-by: hsliuustc0106 <222337142+hsliuustc0106@users.noreply.github.com>
 2026-01-19 04:11:36 +00:00
copilot-swe-agent[bot]andhsliuustc0106 be7d524326 Fix line length linting issue 
Co-authored-by: hsliuustc0106 <222337142+hsliuustc0106@users.noreply.github.com>
 2026-01-19 03:07:31 +00:00
copilot-swe-agent[bot]andhsliuustc0106 c7bdfb2934 Add modalities filtering to non-streaming chat completions 
Co-authored-by: hsliuustc0106 <222337142+hsliuustc0106@users.noreply.github.com>
 2026-01-19 03:04:26 +00:00
copilot-swe-agent[bot] b8c080d848 Initial plan 2026-01-19 03:01:42 +00:00
汪志鹏 36c28767cd [Core]Add GPU Diffusion Runner (#822) 
Signed-off-by: princepride <wangzhipeng628@gmail.com>
 2026-01-17 19:49:13 +08:00
John Liu BUAA bb24e07a7a [Perf] Optimize the Qwen2.5-Omni Model thinker-to-talker-proj with nn.Linear (#825) 
Signed-off-by: John Liu BUAA <liukecheng97@gmail.com>
 2026-01-17 19:47:11 +08:00
汪志鹏 0888520875 [Bagel] Support Cache-Dit (#736) 
Signed-off-by: princepride <wangzhipeng628@gmail.com>
 2026-01-17 15:21:49 +08:00
wangyuandwangyu31577 4e23bfff9e [Test] Add full test for Qwen3-Omni-30B-A3B-Instruct (#720) 
Signed-off-by: wangyu31577 <wangyu31577@hundsun.com>
Co-authored-by: wangyu31577 <wangyu31577@hundsun.com>
 2026-01-16 21:16:43 +08:00
Yueqian LinandHongsheng Liu 3fb6adc40a [bugfix] use unipc scheduler for Wan 2.2 (#804) 
Signed-off-by: linyueqian <linyueqian@outlook.com>
Signed-off-by: Hongsheng Liu <liuhongsheng4@huawei.com>
Co-authored-by: Hongsheng Liu <liuhongsheng4@huawei.com>
 2026-01-16 12:31:25 +00:00
WeiQing Chen a7f9926013 [Model] add flux2 klein (#809) 
Signed-off-by: David Chen <530634352@qq.com>
 2026-01-16 16:32:28 +08:00
Didan Deng 43ec7f701b [Feature] Flash Attention to Support Attention Mask (#760) 
Signed-off-by: Didan Deng <33117903+wtomin@users.noreply.github.com>
 2026-01-16 11:01:11 +08:00
Peiqi Yin 35f994e255 [BugFix] Qwen2.5-omni supress end token and won't stop (#773) 
Signed-off-by: yinpeiqi <yinpeiqi809@gmail.com>
 2026-01-16 00:41:59 +08:00
Ziming Huang b5cf8806f0 [BugFix] Fix incorrect mrope positions under cuda graph (#803) 
Signed-off-by: ZeldaHuang <hzm414167@alibaba-inc.com>
 2026-01-16 03:24:21 +13:00
WeiQing Chen 4203bfa350 [Misc] add wechat group and star history on README (#801) 
Signed-off-by: David Chen <530634352@qq.com>
 2026-01-16 03:22:16 +13:00
Yueqian Lin aaea77a180 [bugfix] Fix Wan2.2 I2V warmup failure by adding support_image_input attribute (#791) 
Signed-off-by: linyueqian <linyueqian@outlook.com>
 2026-01-15 13:14:34 +00:00
Samit 2d5faf3634 [Bugfix] Fix generation artifacts of Qwen-Image-Edit-2511 and update pipeline DiT param parsing (#776) 
Signed-off-by: samithuang <285365963@qq.com>
 2026-01-15 11:18:28 +08:00
Alicia 314bb4e324 [Doc] Fix mkdocs. (#785) 
Signed-off-by: Alicia <115451386+congw729@users.noreply.github.com>
 2026-01-15 10:38:27 +08:00
Alicia 1fe64e85a6 [CI] Add pytest markers in config files. (#719) 
Signed-off-by: Alicia <115451386+congw729@users.noreply.github.com>
 2026-01-14 20:35:26 +08:00
Yuhan Liu e213bdd0a4 [BugFix]Remove duplicate error handling for request results (#781) 
Signed-off-by: Yuhan Liu <30294295+liuyuhanalex@users.noreply.github.com>
 2026-01-14 20:08:23 +08:00
Ziming Huang 1444e1fc45 [Feature] Support Qwen3 Omni talker mtp batch inference (#722) 
Signed-off-by: ZeldaHuang <hzm414167@alibaba-inc.com>
 2026-01-14 19:07:31 +08:00
catcat e9a1beec77 [Bugfix] Fix stable diffusion3 compatibility error (#772) 
Signed-off-by: iwzbi <wzbi@zju.edu.cn>
 2026-01-13 21:12:32 +08:00
Jiangyun Zhu d7cd00e26c [Doc] refactor diffusion doc (#753) 
Signed-off-by: zjy0516 <riverclouds.zhu@qq.com>
 2026-01-13 19:21:24 +08:00
Sihyeon Jang f7611198ae [Feat] add vllm-omni version collection (#740) 
Signed-off-by: Sihyeon Jang <sihyeon.jang@navercorp.com>
 2026-01-13 17:43:55 +08:00
XU Mingshi 8e20b33b38 [Bugfix] Raise ValueError when joint_strategy='rear' and causal=True in Ring Attention (#767) 
Signed-off-by: XU Mingshi <91017482+mxuax@users.noreply.github.com>
Signed-off-by: mxuax <mxuax@connect.ust.hk>
 2026-01-13 17:13:31 +08:00
JustQJ 9cdf592782 [Misc] Enable tensor_parallel_size argument with online serving cmd (#761) 
Signed-off-by: TangPeng <85704592@qq.com>
 2026-01-13 15:55:25 +08:00
Chenguang Zheng 7e14c943d5 [Misc] Group omni arguments into OmniConfig section (#744) 
Signed-off-by: Chenguang ZHENG <645327136@qq.com>
Signed-off-by: ZEHNG Chenguang <645327136@qq.com>
Signed-off-by: ZHENG Chenguang <645327136@qq.com>
 2026-01-12 23:20:20 +08:00
Prajwal Aandzjy0516 779f5987d9 [feature] cpu offloading support for diffusion (#497) 
Signed-off-by: zjy0516 <riverclouds.zhu@qq.com>
Signed-off-by: Prajwal A <prajwalanagani@gmail.com>
Co-authored-by: zjy0516 <riverclouds.zhu@qq.com>
 2026-01-12 10:41:18 +00:00
Zhou Taichangandroot 2e13bddf87 [Debug] Clean code in Qwen 3 Omni and add warning for talker temperature. (#688) 
Signed-off-by: tzhouam <tzhouam@connect.ust.hk>
Co-authored-by: root <root@hk01dgx028.cm.cluster>
 2026-01-12 23:13:45 +13:00
XU Mingshi 8c7f704c23 [Diffusion][Feature] Implement SP support in LongCatImageTransformer (#721) 
Signed-off-by: XU Mingshi <91017482+mxuax@users.noreply.github.com>
Signed-off-by: mxuax <mxuax@connect.ust.hk>
 2026-01-12 00:22:50 -08:00
Sihao LiandHongsheng Liu 2c92b63cce fix:#issue 432 (#517) 
Signed-off-by: GG-li <3226868735@qq.com>
Signed-off-by: Sihao Li <111170255+GG-li@users.noreply.github.com>
Co-authored-by: Hongsheng Liu <liuhongsheng4@huawei.com>
 2026-01-12 15:38:47 +08:00
Hongsheng Liu 61ddeef9f9 [misc] fix rfc template (#748) 
Signed-off-by: hsliu <liuhongsheng4@huawei.com>
Signed-off-by: Hongsheng Liu <liuhongsheng4@huawei.com>
 2026-01-12 15:37:59 +08:00
wangyuandwangyu31577 19887997eb Add description of code version for bug report (#745) 
Signed-off-by: wangyu31577 <wangyu31577@hundsun.com>
Co-authored-by: wangyu31577 <wangyu31577@hundsun.com>
 2026-01-12 15:35:13 +08:00
Hongsheng Liu 36c7b0148f [Docs] update design doc templated in RFC (#746) 
Signed-off-by: hsliu <liuhongsheng4@huawei.com>
 2026-01-12 14:59:06 +08:00
TJian f25cd37f93 [ROCm] [CI] Add More Tests (#542) 
Signed-off-by: tjtanaa <tunjian.tan@embeddedllm.com>
 2026-01-12 14:06:31 +08:00
Junhong Liu d0836d8459 [Bugfix] Fix multi-audio input shape alignment for Qwen3-Omni Thinker (#697) 
Signed-off-by: Junhong Liu <98734602+LJH-LBJ@users.noreply.github.com>
 2026-01-12 13:16:29 +08:00
dongbo910220andJiangyun Zhu 67c0a17a0e [Feat] Enable DiT tensor parallel for Diffusion Pipeline(Z-Image) (#735) 
Signed-off-by: dongbo910220 <1275604947@qq.com>
Signed-off-by: dongbo910220 <32610838+dongbo910220@users.noreply.github.com>
Signed-off-by: Jiangyun Zhu <riverclouds.zhu@qq.com>
Co-authored-by: Jiangyun Zhu <riverclouds.zhu@qq.com>
 2026-01-12 11:00:42 +08:00
Canlin Guo dde96c2468 [Perf][Qwen3-Omni] Fuse QKV linear and gate_up proj (#734) 
Signed-off-by: gcanlin <canlinguosdu@gmail.com>
 2026-01-11 11:06:24 +08:00
汪志鹏 0ec2da08de [BugFix] Fix assuming all stage model have talker (#730) 
Signed-off-by: princepride <wangzhipeng628@gmail.com>
 2026-01-11 01:36:37 +13:00
JohnJanHongsheng LiuWeiQing Chen
 		f3741ac2b9 [Doc]: update vllm serve param and base64 data truncation (#718) 
Signed-off-by: wuzhongjian <wuzhongjian_yewu@cmss.chinamobile.com>
Signed-off-by: Hongsheng Liu <liuhongsheng4@huawei.com>
Co-authored-by: Hongsheng Liu <liuhongsheng4@huawei.com>
Co-authored-by: WeiQing Chen <40507679+david6666666@users.noreply.github.com>
 2026-01-11 00:02:32 +13:00
Canlin GuoandIsotr0py 9c2f7467a0 [Perf] Use vLLM's SharedFusedMoE in Qwen3-Omni (#560) 
Signed-off-by: gcanlin <canlinguosdu@gmail.com>
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
Co-authored-by: Isotr0py <mozf@mail2.sysu.edu.cn>
 2026-01-10 04:30:06 +00:00
Hongsheng Liu 65b55d1c83 update design doc (#711) 
Signed-off-by: hsliu <liuhongsheng4@huawei.com>
 2026-01-09 13:30:17 +08:00
TsuKKIandasumaska f26219162b [Benchmark] DiT Model Benchmark under Mixed Workloads (#529) 
Signed-off-by: asumaska <asumasaka@noreply.gitcode.com>
Co-authored-by: asumaska <asumasaka@noreply.gitcode.com>
 2026-01-09 12:36:53 +08:00
Ziming Huang 978e3ad7d4 [Feature] Support Qwen3 Omni talker cudagraph (#669) 
Signed-off-by: ZeldaHuang <hzm414167@alibaba-inc.com>
 2026-01-09 11:45:58 +08:00
dongbo910220 1467b3e2c7 [BugFix] Remove duplicate width assignment in SD3 pipeline (#708) 
Signed-off-by: dongbo910220 <1275604947@qq.com>
 2026-01-09 08:34:17 +08:00
Jiangyun Zhu c784b3e07d [core] add torch compile for diffusion (#684) 
Signed-off-by: zjy0516 <riverclouds.zhu@qq.com>
 2026-01-09 04:05:44 +13:00
Ziming Huang cddcc400d3 [BugFix] Fix stage engine outputs mismatch bug in online batching (#691) 
Signed-off-by: ZeldaHuang <hzm414167@alibaba-inc.com>
 2026-01-09 04:01:57 +13:00
Yuanheng Zhao 4c72b67444 [BugFix] Explicitly release file locks during stage worker init (#703) 
Signed-off-by: yuanheng <jonathan.zhaoyh@gmail.com>
 2026-01-08 20:35:00 +08:00
Junhong Liu bb13467adc [Bug fix] fix e2e_total_tokens and e2e_total_time_ms (#648) 
Signed-off-by: Junhong Liu <98734602+LJH-LBJ@users.noreply.github.com>
 2026-01-08 20:25:04 +08:00
dsinghvi 8c1259383c omni entrypoint support tokenizer arg (#572) 
Signed-off-by: Divyansh Singhvi <divyanshsinghvi@gmail.com>
 2026-01-08 15:44:14 +08:00
Didan Deng 4df794bdaa [Bugfix] Solve Ulysses-SP sequence length not divisible by SP degree (using padding and attention mask) (#672) 
Signed-off-by: Didan Deng <33117903+wtomin@users.noreply.github.com>
 2026-01-08 00:02:22 +08:00