Alex Brooks and lengrongfu
6c57ab7ac2
[Misc] Warn When vLLM / vLLM-Omni Have Mismatched Versions ( #2691 )
...
Signed-off-by: Alex Brooks <albrooks@redhat.com >
Co-authored-by: lengrongfu <lenronfu@gmail.com >
2026-04-17 20:15:16 +08:00
bjf-frz
18ac679763
[Refactor] refactor wan2.2 diffuse && add ut ( #2672 )
...
Signed-off-by: bjf-frz <frz123db@gmail.com >
2026-04-17 20:13:49 +08:00
6b7be88cd5
Fix NoneType error of outputs ( #2315 )
...
Signed-off-by: Mike_Qiu <qiudayu.qdy@antgroup.com >
Co-authored-by: Mike_Qiu <qiudayu.qdy@antgroup.com >
Co-authored-by: Hongsheng Liu <liuhongsheng4@huawei.com >
2026-04-17 20:10:24 +08:00
Lancer and Hongsheng Liu
cf75ae6ed5
[Bugfix] Fix image quality in /v1/images/generations for multi-stage pipeline ( #2267 )
...
Signed-off-by: Lancer <maruixiang6688@gmail.com >
Co-authored-by: Hongsheng Liu <liuhongsheng4@huawei.com >
2026-04-17 20:09:41 +08:00
Zhou Taichang
edb4f2fdcf
[Test] Add ModelRunner V2 with Qwen3-TTS Base E2E Test to CI pipeline ( #2321 )
...
Signed-off-by: tzhouam <tzhouam@connect.ust.hk >
2026-04-17 20:08:23 +08:00
WeiQing Chen and Gao Han
f658bcb0bf
[Bugfix] Limit Qwen-Image-Edit-2511 input image count ( #2840 )
...
Signed-off-by: david6666666 <530634352@qq.com >
Co-authored-by: Gao Han <hgaoaf@connect.ust.hk >
2026-04-17 20:07:30 +08:00
WeiQing Chen
b7f239810d
[Bugfix] Fix RIFE device selection for CPU-transported videos ( #2876 )
...
Signed-off-by: david6666666 <530634352@qq.com >
2026-04-17 19:19:34 +08:00
Yuanheng Zhao
c0ccbb872f
[Model] Add Ming-flash-omni-2.0 Thinker Stage ( #1822 )
...
Signed-off-by: yuanheng <jonathan.zhaoyh@gmail.com >
Signed-off-by: Yuanheng Zhao <jonathan.zhaoyh@gmail.com >
2026-04-17 19:10:02 +08:00
Bingyu (Spencer) Liu
a5a4998ece
[Feature] Support Prefill-Decode disaggregation via vLLM KV transfer ( #2220 )
...
Signed-off-by: LiuBingyu <liubingyu62@gmail.com >
2026-04-17 18:43:17 +08:00
amy-why-3459
b88d3ce75e
[BugFix] Fixing occasional engine crashes caused by abort requests ( #2871 )
...
Signed-off-by: amy-why-3459 <wuhaiyan17@huawei.com >
2026-04-17 09:15:35 +00:00
wangyu
a3ecde98e0
[CI][Bugfix] Fix the error in generating the performance data table and add a fallback mechanism that prevents the result file from being generated when test case execution fails. ( #2839 )
...
Signed-off-by: wangyu <410167048@qq.com >
Signed-off-by: wangyu <53896905+yenuo26@users.noreply.github.com >
2026-04-17 16:33:46 +08:00
Peiqi Yin
bbd6a443ed
[Core] Refactor CFG companion tracker and use in Orchestrator ( #2623 )
...
Signed-off-by: yinpe <11810305@mail.sustech.edu.cn >
2026-04-17 08:13:00 +00:00
Alex Brooks
d463978098
[Model] Support Flux1 Schnell ( #2528 )
...
Signed-off-by: Alex Brooks <albrooks@redhat.com >
2026-04-17 16:02:30 +08:00
WeiQing Chen
12378824fb
[Bugfix] Preserve default diffusion sampling params in default stage ( #2780 )
...
Signed-off-by: David Chen <530634352@qq.com >
2026-04-17 12:07:10 +08:00
WeiQing Chen
3079e943de
[Bugfix] enforce max_sequence_length for Qwen-Image and Wan2.2 series before encoding ( #2847 )
...
Signed-off-by: david6666666 <530634352@qq.com >
2026-04-17 10:40:36 +08:00
Nick Cao and Claude
1e3bb36521
[Refactor] Remove sox from dependencies ( #2745 )
...
Signed-off-by: Nick Cao <ncao@redhat.com >
Co-authored-by: Claude <noreply@anthropic.com >
2026-04-17 01:24:47 +00:00
Lancer and Hongsheng Liu
bf64a6b841
[Feature] Flux2 klein inpaint ( #1180 )
...
Signed-off-by: Lancer <maruixiang6688@gmail.com >
Co-authored-by: Hongsheng Liu <liuhongsheng4@huawei.com >
2026-04-17 07:55:00 +08:00
Zhang Jian
817e32d548
[Quantization] feat: add FP8 for Omnigen2 ( #2441 )
...
Signed-off-by: Zhang <jianmusings@gmail.com >
2026-04-16 23:33:40 +08:00
c3ca5daafb
Feat/Add HunyuanImage-3.0-Instruct ar part support: ( #2713 )
...
Signed-off-by: TaffyOfficial <2324465096@qq.com >
Co-authored-by: TaffyOfficial <2324465096@qq.com >
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com >
2026-04-16 23:15:25 +08:00
Sy03
7d64a7c996
[BugFix][VoxCPM2]: split multichar Chinese tokens to match training tokenization ( #2832 )
...
Signed-off-by: Sy03 <1370724210@qq.com >
2026-04-16 10:07:48 -05:00
2ec91d4dfd
[FIX] Preserve YAML default stop words when request sends empty list ( #2855 )
...
Signed-off-by: Mike_Qiu <qiudayu.qdy@antgroup.com >
Co-authored-by: Mike_Qiu <qiudayu.qdy@antgroup.com >
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com >
2026-04-16 14:40:34 +00:00
wangyu
45760d61d2
[Test] Skip tests for known issues in audio and speaker recognition ( #2851 )
2026-04-16 21:27:43 +08:00
Sy03
322620fd57
[Fix][Fish Speech] Remove redundant get_vocab() in control token encoding ( #2842 )
...
Signed-off-by: Sy03 <1370724210@qq.com >
2026-04-16 12:47:39 +00:00
John Liu BUAA
e8658b55d1
[Test] Add performance tests for Qwen-Image-Layered model ( #2807 )
...
Signed-off-by: John Liu BUAA <liukecheng97@gmail.com >
2026-04-16 18:49:59 +08:00
fan2956 and gcanlin
f1cb4ebe4c
[PERF] Wan2.2 support rmsnorm fused op ( #2583 )
...
Signed-off-by: fan2956 <zhoufan53@huawei.com >
Signed-off-by: gcanlin <canlinguosdu@gmail.com >
Co-authored-by: gcanlin <canlinguosdu@gmail.com >
2026-04-16 03:21:34 -07:00
NATURE and Hongsheng Liu
4d816ff1de
[Feature] Bagel: Support tp+cfg parallel using mooncake transfer engine connector ( #2705 )
...
Signed-off-by: natureofnature <wzliu@connect.hku.hk >
Co-authored-by: Hongsheng Liu <liuhongsheng4@huawei.com >
2026-04-16 16:25:13 +08:00
WeiQing Chen
24e61f4d7b
[CI] add qwen image and layered accuracy test ( #2772 )
...
Signed-off-by: david6666666 <530634352@qq.com >
2026-04-16 14:52:53 +08:00
Lancer and Didan Deng
b43c6c6663
[Feat] add GLM-Image SP support ( #1983 )
...
Signed-off-by: Lancer <maruixiang6688@gmail.com >
Signed-off-by: Didan Deng <33117903+wtomin@users.noreply.github.com >
Co-authored-by: Didan Deng <33117903+wtomin@users.noreply.github.com >
2026-04-16 12:20:46 +08:00
Yuanheng Zhao
de5f8a23b2
[Doc][Misc] Update DreamID-Omni Example; Add DreamID-Omni post process function ( #2809 )
...
Signed-off-by: yuanheng <jonathan.zhaoyh@gmail.com >
2026-04-16 03:26:48 +00:00
c83f664fe1
[CI][Perf] Add nightly PR labels, consolidate pipeline, and switch benchmark flag to --test-config-file ( #2816 )
...
Signed-off-by: wangyu <410167048@qq.com >
Co-authored-by: Y. Fisher <yukexiong1@huawei.com >
Co-authored-by: inaniloquentee <inani_@stu.xjtu.edu.cn >
2026-04-16 11:13:41 +08:00
wangyu
880a758b1f
[CI] Skip test_thinker_prefix_caching in tests/e2e/online_serving/test_qwen3_omni.py ( #2836 )
...
Signed-off-by: wangyu <410167048@qq.com >
2026-04-16 10:15:39 +08:00
Didan Deng
e9581137e9
[Perf] Add Performance Test for Qwen-Image Step-Level Execution ( #2707 )
...
Signed-off-by: Didan Deng <33117903+wtomin@users.noreply.github.com >
2026-04-15 17:45:16 +00:00
Alex Brooks
f1e3f03726
[feature] Hidden State Prefix Caching ( #2164 )
...
Signed-off-by: Alex Brooks <albrooks@redhat.com >
2026-04-15 17:01:45 +00:00
Zhang Jian and pjh4993
c6d76d081b
[Bugfix] Fix broken fp8 quantisation on Z-Image-Turbo, Qwen-Image, FLUX.1-dev ( #2795 )
...
Signed-off-by: Zhang <jianmusings@gmail.com >
Co-authored-by: pjh4993 <pjh4993@naver.com >
2026-04-15 14:25:41 +00:00
Y. Fisher
50ae1de7da
[Feature] HunyuanImage3 allow guidance_scale<=1 in DiT stage ( #2762 )
...
Signed-off-by: KexiongYu <yukexiong1@huawei.com >
2026-04-15 15:54:38 +08:00
Juan Pablo Zuluaga and Hongsheng Liu
82f8c93343
[Feat][Qwen3-Omni] Shared code predictor module for Qwen3-TTS and Qwen3-Omni ( #2375 )
...
Signed-off-by: JuanPZuluaga <juanz9312@gmail.com >
Co-authored-by: Hongsheng Liu <liuhongsheng4@huawei.com >
2026-04-15 15:14:57 +08:00
4bf4c63147
Add voxcpm model support. ( #2467 )
...
Signed-off-by: Celeste-jq <591998922@qq.com >
Signed-off-by: lyj-jjj <liuyingjun5@huawei.com >
Signed-off-by: IsleOfDawnlight <stellamou@qq.com >
Signed-off-by: Yueqian Lin <linyueqian@outlook.com >
Co-authored-by: Celeste-jq <591998922@qq.com >
Co-authored-by: lyj-jjj <liuyingjun5@huawei.com >
Co-authored-by: Yueqian Lin <linyueqian@outlook.com >
2026-04-15 15:04:58 +08:00
Sy03
38d5f2d530
[Perf] VoxCPM2: Speedup by manual CUDA Graph capture for scaffold/residual forward ( #2803 )
...
Signed-off-by: Sy03 <1370724210@qq.com >
2026-04-15 14:09:22 +08:00
Zeyu Huang | 黃澤宇
133e2f9706
[hotfix] API connection error in CI ( #2810 )
2026-04-15 14:08:00 +08:00
fan2956 and Canlin Guo
2dff2d7c74
[PERF] Wan2.2 support adalayernorm fused op ( #2585 )
...
Signed-off-by: fan2956 <zhoufan53@huawei.com >
Co-authored-by: Canlin Guo <canlinguosdu@gmail.com >
2026-04-15 14:02:41 +08:00
TaffyOfficial
1ad726f495
[Fix] HunyuanImage-3.0: unify naming hunyuan_image_3 → hunyuan_image3 ( #2712 )
2026-04-14 21:53:47 -07:00
WeiQing Chen
6c6551dff8
[Feature] feat: add video frame interpolation postprocess ( #2555 )
...
Signed-off-by: David Chen <530634352@qq.com >
2026-04-15 11:52:20 +08:00
Samit
61a3cbdff5
[Perf] Eliminate Hop 3 IPC overhead for single-stage diffusion via inline execution ( #2736 )
...
Signed-off-by: samithuang <285365963@qq.com >
Signed-off-by: Samit <285365963@qq.com >
2026-04-15 11:46:06 +08:00
Zeyu Huang | 黃澤宇
0d020739a7
[CI] qwen image edit L4 accuracy test ( #2761 )
2026-04-15 11:13:11 +08:00
amy-why-3459
227bab3038
[Benchmark]Omni-modality model accuracy benchmark(Daily-Omni & seed-tts-eval) ( #2558 )
...
Signed-off-by: amy-why-3459 <wuhaiyan17@huawei.com >
2026-04-15 11:07:35 +08:00
wangyu
a782ae4780
[Perf] Enhance benchmark script to support baseline thresholds and proved result handling ( #2789 )
2026-04-15 10:28:28 +08:00
Yueqian Lin
02e5dc747d
[Perf] VoxCPM2: streaming VAE + compile optimization (45% RTF reduction) ( #2758 )
...
Signed-off-by: Yueqian Lin <linyueqian@outlook.com >
2026-04-15 09:58:25 +08:00
rongfu.leng
9e46a79c17
[Bugfix] Handle numpy array outputs when generate image ( #1680 )
...
Signed-off-by: rongfu.leng <lenronfu@gmail.com >
2026-04-15 08:52:32 +08:00
Yueqian Lin
033e4041be
fix(moss-tts-nano): remove unused voice variable (ruff F841)
...
The upstream inference_stream() API has no voice/spk parameter;
voice preset selection is not yet wired into the call. Remove the
dead assignment to silence ruff F841.
Signed-off-by: Yueqian Lin <linyueqian@outlook.com >
2026-04-14 13:11:43 -04:00
Yueqian Lin
38aaaebeda
feat(moss-tts-nano): enable progressive streaming via AR runner
...
Switch from GPUGenerationWorker to AR worker with OmniARScheduler,
using the VoxCPM-style generator pattern for streaming:
- inference_stream() stored per-request in _stream_gens dict
- Each forward() call yields one audio chunk via next(generator)
- compute_logits() emits EOS only when last chunk is yielded
- AR scheduler loops until EOS, enabling progressive audio output
TTFP reduced from ~3.1s to ~0.11s (30x improvement).
Signed-off-by: Yueqian Lin <linyueqian@outlook.com >
2026-04-14 13:06:44 -04:00