 Paul ZhangandWentao Ye
|
e7acb20076
|
[Feature] Batch invariant torch.compile (#27660)
Signed-off-by: PaulZhang12 <paulzhan@fb.com>
Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
|
2025-10-30 13:11:29 -07:00 |
|
|
Tyler Michael SmithandNick Hill
|
ab98f6556f
|
[Bugfix] Fix 2 precommit issues - (mamba_block_size, kv_cache_config) (#27811)
Signed-off-by: Tyler Michael Smith <tlrmchlsmth@gmail.com>
Signed-off-by: Tyler Michael Smith <tysmith@redhat.com>
Co-authored-by: Nick Hill <nhill@redhat.com>
|
2025-10-30 11:52:18 -07:00 |
|
|
Roger Meier
|
2918c1b49c
|
[Model] Use the same fused_moe configs for all H200 devices (#23642)
Signed-off-by: Roger Meier <r.meier@siemens.com>
|
2025-10-30 17:36:56 +00:00 |
|
|
Mengqing Cao
|
1004205795
|
[MTP] Refactor mtp predictor to avoid d2h operation (#27643)
Signed-off-by: MengqingCao <cmq0113@163.com>
|
2025-10-30 17:27:39 +00:00 |
|
|
Varun Sundar RabindranathandVarun Sundar Rabindranath
|
e5e076cad7
|
[BugFix] Stopgap - Flashinfer Autotuner + GPT-OSS + DP/TP (#27762)
Signed-off-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
Co-authored-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
|
2025-10-30 08:24:31 -07:00 |
|
|
Li, Jiang
|
eebf00cb0c
|
[Bugfix][CPU] Fix MRoPE dispatch on the CPU backend (#27800)
Signed-off-by: jiang1.li <jiang1.li@intel.com>
|
2025-10-30 15:12:05 +00:00 |
|
|
|
9956aae4ea
|
[Model][Ouro] Support Ouro Model (#27794)
Signed-off-by: yinfan.1024 <yinfan.1024@bytedance.com>
Signed-off-by: youkaichao <youkaichao@gmail.com>
Co-authored-by: yinfan.1024 <yinfan.1024@bytedance.com>
Co-authored-by: youkaichao <youkaichao@gmail.com>
Co-authored-by: Jee Jee Li <pandaleefree@gmail.com>
|
2025-10-30 22:34:41 +08:00 |
|
|
Zhiyuan Li
|
4e68cc9b6a
|
[Model] Introduce Kimi Linear to vLLM (#27809)
Signed-off-by: lizhiyuan <lizhiyuan@moonshot.cn>
Signed-off-by: Zhiyuan Li <uniartisan2017@gmail.com>
|
2025-10-30 21:02:27 +08:00 |
|
|
|
4464723f22
|
[Frontend][Doc][5/N] Improve all pooling task | Polish encode (pooling) api & Document. (#25524)
Signed-off-by: wang.yuqi <noooop@126.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>
|
2025-10-30 12:13:05 +00:00 |
|
|
Zhewen LiandRoger Wang
|
e806178d2a
|
[BugFix][VL] Fix FA selection on Qwen2.5-VL (#27790)
Signed-off-by: zhewenli <zhewenli@meta.com>
Co-authored-by: Roger Wang <hey@rogerw.io>
|
2025-10-30 07:54:44 +00:00 |
|
|
Bram Wastiandgemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
|
ded8ada86a
|
Add more dims for batch invariant shims (#27489)
Signed-off-by: Bram Wasti <bwasti@meta.com>
Signed-off-by: Bram Wasti <bwasti@fb.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
|
2025-10-30 05:28:45 +00:00 |
|
|
|
17d055f527
|
[Feat] Adds runai distributed streamer (#27230)
Signed-off-by: bbartels <benjamin@bartels.dev>
Signed-off-by: Benjamin Bartels <benjamin@bartels.dev>
Co-authored-by: omer-dayan <omdayan@nvidia.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
|
2025-10-29 21:09:10 -07:00 |
|
|
Yan Ma
|
b798e39f93
|
[XPU][bugfix] fix rope for llama4 and deepseek (#25145)
Signed-off-by: Yan Ma <yan.ma@intel.com>
|
2025-10-30 09:43:13 +08:00 |
|
|
Chenheli HuaandRoger Wang
|
48eb8eba58
|
[Temp fix] Disable torch.compile for Qwen2.5 VL's VisionBlock temporarily. (#27760)
Signed-off-by: Chenheli Hua <huachenheli@outlook.com>
Signed-off-by: Roger Wang <hey@rogerw.io>
Co-authored-by: Roger Wang <hey@rogerw.io>
|
2025-10-29 23:17:48 +00:00 |
|
|
Wentao Ye
|
b5d90f7400
|
[Bug] Fix DBO IMA issue for DeepEPHT (#27666)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
|
2025-10-29 16:28:27 -04:00 |
|
|
Wentao Ye
|
fcb1d570bb
|
[Bug] Fix DeepEP low latency assert self.batched_router_logits.size(-1) == full_router_logits.size(-1) Bug (#27682)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
|
2025-10-29 14:50:39 -04:00 |
|
|
JartXandLukas Geiger
|
7568a282b9
|
[FIXBUG] Qwen3VL hallucinations without Contiguous on Torch.SDPA (#27744)
Signed-off-by: JartX <sagformas@epdcenter.es>
Co-authored-by: Lukas Geiger <lukas.geiger94@gmail.com>
|
2025-10-29 16:55:35 +00:00 |
|
|
Roger Youngandxuebi
|
d6704dd099
|
Fix MiniMax-M2 rmsnorm precision and remove useless code (#27627)
Signed-off-by: xuebi <xuebi@minimaxi.com>
Co-authored-by: xuebi <xuebi@minimaxi.com>
|
2025-10-29 21:01:05 +08:00 |
|
|
Jiangyun Zhu
|
8df98c2161
|
[perf] Enable concurrent execution of "shared_experts" and "selected_experts" in qwen3-next (#27578)
Signed-off-by: zjy0516 <riverclouds.zhu@qq.com>
|
2025-10-29 08:12:54 +00:00 |
|
|
Zhewen Li
|
8b62495076
|
[Bugfix] Fix non-contiguous tensor error in rocm_unquantized_gemm_impl (#27605)
Signed-off-by: zhewenli <zhewenli@meta.com>
|
2025-10-29 00:00:15 -07:00 |
|
|
Lukas GeigerandRoger Wang
|
0d8161b075
|
[Model] Fix Qwen3VL and Qwen3Omni after torch.compile changes (#27705)
Signed-off-by: Lukas Geiger <lukas.geiger94@gmail.com>
Signed-off-by: Roger Wang <hey@rogerw.io>
Co-authored-by: Roger Wang <hey@rogerw.io>
|
2025-10-29 05:28:20 +00:00 |
|
|
Lucas Kabela
|
94666612a9
|
[Misc][qwen2_5_vl][torch.compile] Enable supports_torch_compile on generic nn.Module and demonstrate speedup on Qwen Vision model (#23207)
Signed-off-by: Lucas Kabela <lucaskabela@meta.com>
Signed-off-by: Lucas Kabela <lucasakabela@gmail.com>
|
2025-10-28 22:36:43 +00:00 |
|
|
Wentao Ye
|
6afc28a9ba
|
[Test] Batch Invariant: Unit test using parameterized backend (#27478)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
|
2025-10-28 13:51:35 -07:00 |
|
|
Zhiyuan Li
|
e88bdd60d9
|
[FLA] Introduce Kimi Delta Attention(KDA) to VLLM (#27654)
Signed-off-by: lizhiyuan <lizhiyuan@moonshot.cn>
|
2025-10-28 22:56:28 +08:00 |
|
|
Asaf Joseph Gardin
|
05181cc57f
|
[Hybrid] Add mamba_block_size to Engine Args (#27289)
Signed-off-by: asafg <39553475+Josephasafg@users.noreply.github.com>
|
2025-10-28 12:54:24 +00:00 |
|
|
Wentao YeandCyrus Leung
|
0484b64248
|
[Bug] Fix shape issue for eplb expert weights (#27589)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
|
2025-10-28 20:44:05 +08:00 |
|
|
Matthew Bonanni
|
44b5ce956d
|
[Bugfix] In LongRoPE, decide short vs long based on max_model_len (#27431)
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
|
2025-10-28 12:00:56 +00:00 |
|
|
Li, JiangandIsotr0py
|
d34f5fe939
|
[Bugfix][CPU] Fallback oneDNN linear to torch linear to fix half gemm support on legecy platforms (#27526)
Signed-off-by: jiang1.li <jiang1.li@intel.com>
Co-authored-by: Isotr0py <mozf@mail2.sysu.edu.cn>
|
2025-10-27 23:25:44 -07:00 |
|
|
Eric Yue
|
bdb01a38fe
|
[Hardware][AMD][Model] Triton MoE tuning configs for GLM-4.6 for MI300X (#27323)
Signed-off-by: minatoaquaMK2 <jiacheng.yue@foxmail.com>
|
2025-10-27 22:58:06 -07:00 |
|
|
tingtinggithub
|
23ad820553
|
fixing mm placeholder replacement issue with gemma3 (#27538)
Signed-off-by: tingtingtang1992 <streamttt@gmail.com>
|
2025-10-27 14:34:01 +00:00 |
|
|
Varun Sundar RabindranathandVarun Sundar Rabindranath
|
5d3be3ba4c
|
[Bugfix][LoRA][FusedMoE] Select MxFP4 Backend based on LoRA Enablement (#27487)
Signed-off-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
Co-authored-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
|
2025-10-27 07:32:50 -07:00 |
|
|
Yu Jiaqi
|
4f882be4a0
|
[Model] Siglip2 Model Support (#27566)
Signed-off-by: piood <2477084691@qq.com>
|
2025-10-27 06:57:37 -07:00 |
|
|
Asaf Joseph Gardin
|
9273754222
|
[Hybrid] Added supports_mamba_prefix_caching Protocol (#27339)
Signed-off-by: asafg <39553475+Josephasafg@users.noreply.github.com>
|
2025-10-27 13:05:20 +00:00 |
|
|
Cyrus Leung
|
7c2bdb83dc
|
[Misc] Clean up utils (#27552)
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
|
2025-10-27 09:05:40 +00:00 |
|
|
|
9932ed6a83
|
[Kernel] Adding split_K implementation for fused_moe_lora (#27291)
Signed-off-by: Danielle Robinson <dmmaddix@amazon.com>
Signed-off-by: Danielle Robinson <dcmaddix@gmail.com>
Co-authored-by: Danielle Robinson <dmmaddix@amazon.com>
Co-authored-by: Jee Jee Li <pandaleefree@gmail.com>
|
2025-10-27 02:05:24 -07:00 |
|
|
Jee Jee LiandRoger Wang
|
2d631d28c6
|
[Doc] Slight improvement to M2 and beyond (#27554)
Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
Co-authored-by: Roger Wang <hey@rogerw.io>
|
2025-10-27 09:02:10 +00:00 |
|
|
Cyrus Leung
|
cbd5e07a51
|
[Model] Use merge_by_field_config for MM models (Qwen series) (#27546)
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
|
2025-10-27 05:38:05 +00:00 |
|
|
CSWYF3634076
|
63b22e0dbb
|
[Model][Bugfix] fix ernie45 moe 300B SharedFusedMoE output tuple (#27316)
Signed-off-by: wangyafeng <wangyafeng@baidu.com>
|
2025-10-26 20:53:31 -07:00 |
|
|
Roger Youngandxuebi
|
5980604c44
|
Fix MiniMax-M2 copyright (#27537)
Signed-off-by: xuebi <xuebi@minimaxi.com>
Co-authored-by: xuebi <xuebi@minimaxi.com>
|
2025-10-27 03:29:51 +00:00 |
|
|
Roger Youngandxuebi
|
720af6ab79
|
[Model][MiniMax-M2] Support MiniMax-M2 Model (#27535)
Signed-off-by: xuebi <xuebi@minimaxi.com>
Co-authored-by: xuebi <xuebi@minimaxi.com>
|
2025-10-27 00:59:11 +08:00 |
|
|
Yeshwanth N
|
71b1c8b667
|
[Chore]:Extract math and argparse utilities to separate modules (#27188)
Signed-off-by: Yeshwanth Surya <yeshsurya@gmail.com>
Signed-off-by: Yeshwanth N <yeshsurya@gmail.com>
Signed-off-by: yeshsurya <yeshsurya@gmail.com>
|
2025-10-26 04:03:32 -07:00 |
|
|
JartXandtjtanaa
|
65d2cf9511
|
[BUGFIX][ROCM] ViT FlashAttention on ROCm (no GFX9) and contiguous on qwen3vl ROCm TORCH_SDPA (#27190)
Signed-off-by: JartX <sagformas@epdcenter.es>
Co-authored-by: tjtanaa <tunjian.tan@embeddedllm.com>
|
2025-10-26 15:08:52 +08:00 |
|
|
Cyrus LeungandIsotr0py
|
66a168a197
|
[CI/Build] Refactor processing tests (#27470)
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
Co-authored-by: Isotr0py <mozf@mail2.sysu.edu.cn>
|
2025-10-25 16:14:30 +00:00 |
|
|
Varun Sundar RabindranathandVarun Sundar Rabindranath
|
269c4db0a4
|
[Misc][DP] Guard mxfp4 implementation selection (#27484)
Signed-off-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
Co-authored-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
|
2025-10-24 23:29:24 +00:00 |
|
|
Wentao Ye
|
52efc34ebf
|
[Log] Optimize Startup Log (#26740)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
|
2025-10-24 19:27:04 -04:00 |
|
|
Isotr0py
|
acc78aeb88
|
[Bugfix] Fix interns1-vit qk norm code path (#27480)
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
|
2025-10-24 17:43:45 +00:00 |
|
|
|
284cc92275
|
[MISC] cudagraph_capture_sizes related improvements (#26016)
Signed-off-by: fhl <2410591650@qq.com>
Signed-off-by: fhl2000 <63384265+fhl2000@users.noreply.github.com>
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
|
2025-10-24 05:11:05 -07:00 |
|
|
Isotr0pyandRoger Wang
|
42efe609ba
|
[MM][Bugfix] Replace PatchEmbed's conv3d to linear layer (#27418)
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
Co-authored-by: Roger Wang <hey@rogerw.io>
|
2025-10-24 07:32:47 +00:00 |
|
|
Xiangyu Li
|
5cc6bddb6e
|
[Kernel] Add GPTQv2 format support for low-bit or asymmetric quantization, by adapting gptq_gemm (#26092)
|
2025-10-23 23:26:13 -04:00 |
|
|
Harry Mellor
|
1f9460c4c1
|
Fix pooling adapters for Transformers backend (#27338)
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
|
2025-10-23 20:23:55 -07:00 |
|