|
b2f685834a
|
[Hy4] support Hy4-preview model (#54160)
Signed-off-by: chengvjiang <chengvjiang@tencent.com>
Signed-off-by: Jee Jee Li <jeejeelee@inferact.ai>
Signed-off-by: Bugen Zhao <i@bugenzhao.com>
Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
Co-authored-by: chengvjiang <chengvjiang@tencent.com>
Co-authored-by: Jee Jee Li <jeejeelee@inferact.ai>
Co-authored-by: Bugen Zhao <i@bugenzhao.com>
Co-authored-by: russellfeng <russellfeng@tencent.com>
Co-authored-by: Jee Jee Li <pandaleefree@gmail.com>
|
2026-08-29 12:07:28 +08:00 |
|
|
Nick Hill
|
6c18a54648
|
[Perf] Avoid h2d copies from non-pinned CPU tensors (#54299)
Signed-off-by: Nick Hill <nickhill123@gmail.com>
|
2026-08-29 12:06:30 +08:00 |
|
|
|
46a83642f6
|
[Perf] Pin CPU tensors before non_blocking H2D in three MM paths (#54292)
Signed-off-by: Keyi Li <likey6688@gmail.com>
Co-authored-by: Keyi Li <likey6688@gmail.com>
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
|
2026-08-28 20:13:40 -07:00 |
|
|
Nick HillandKimi Code
|
70732942c6
|
[Bugfix][ROCm] Pre-allocate wvSplitKrc static workspaces before KV init (#54247)
Signed-off-by: Nick Hill <nickhill123@gmail.com>
Co-authored-by: Kimi Code <noreply@moonshot.ai>
|
2026-08-28 14:22:53 -07:00 |
|
|
xuebwang-amd
|
06569a8696
|
[ROCm][Quantization][MOE] Enable fused shared experts for block-quantized FP8 (#53097)
Signed-off-by: xuebwang-amd <xuebwang@amd.com>
|
2026-08-28 09:23:07 -07:00 |
|
|
maobaolongandCodex
|
6f7df92a8e
|
[CPU][MLA] Fix prefill backend selection so MLA runs end-to-end on CPU (#51471)
Signed-off-by: baoloongmao <baoloongmao@tencent.com>
Co-authored-by: Codex <codex@openai.com>
|
2026-08-28 12:36:55 +08:00 |
|
|
Thien Tran
|
6ec92bcbc8
|
[Kimi-K3] Merge MLA gate into QKV-A projection (#54015)
Signed-off-by: Thien Tran <gau.nernst@yahoo.com.sg>
|
2026-08-27 17:47:18 -07:00 |
|
|
|
7c877062ac
|
[kernel] Integrate FlashInfer BF16 CuTeDSL Low Latency GEMM (#50572)
Signed-off-by: jiahanc <173873397+jiahanc@users.noreply.github.com>
Co-authored-by: OpenAI Codex <codex@openai.com>
Co-authored-by: Jee Jee Li <pandaleefree@gmail.com>
|
2026-08-27 22:29:12 +00:00 |
|
|
Matthew BonanniandOpenAI Codex
|
de9250ac9e
|
[Attention] Enable masked MHA for GLM-5 head dimensions (#53785)
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
Co-authored-by: OpenAI Codex <noreply@openai.com>
|
2026-08-27 21:17:12 +00:00 |
|
|
Chauncey
|
7d5769b2d9
|
[Perf][DSv4] Use native CUDA SwiGLU clamp kernel for Humming MoE (throughput +1.40%) (#53685)
Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
|
2026-08-27 13:52:45 -07:00 |
|
|
Elvir Crnčević
|
674c284dbd
|
Fix Humming MoE activation_output aliasing (#54056)
|
2026-08-27 18:28:38 +00:00 |
|
|
Khushali Desai
|
c7b0467c25
|
buffer size insuffient Dspark sd for FlashInfer MNNVL allreduce (#50932)
|
2026-08-27 11:01:08 -04:00 |
|
|
Robert ShawandRobert Shaw
|
6a5e8f5979
|
[DeepEPv2] Support MXFp8 Activation Scale Dispatch (#51398)
Signed-off-by: Robert Shaw <robshaw@redhat.com>
Co-authored-by: Robert Shaw <robshaw@redhat.com>
|
2026-08-26 12:26:11 +00:00 |
|
|
|
2267d3b112
|
[AttentionBackend][HPC-ops] update hpc rope norm to support stride kv cache (#53705)
Signed-off-by: chengvjiang <chengvjiang@tencent.com>
Co-authored-by: chengvjiang <chengvjiang@tencent.com>
Co-authored-by: Yongye Zhu <zyy1102000@gmail.com>
|
2026-08-26 08:20:28 +00:00 |
|
|
|
044b05220e
|
[Kernel][Perf] Tune fused_moe FP8 config for Qwen3.5 on L40S (+7%) (#53819)
Signed-off-by: nizhang1 <nizhang1@coupang.com>
Co-authored-by: nizhang1 <nizhang1@coupang.com>
Co-authored-by: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
|
2026-08-26 04:02:26 +00:00 |
|
|
Wentao Yeandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
a5e0045433
|
[Refactor] Remove dead code quantization 2 (#52821)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
2026-08-25 20:28:44 -07:00 |
|
|
|
ca29cd48ba
|
[Quantization][Humming] Support MXFP4 weight + block-FP8 activation for MoE (#51332)
Signed-off-by: Elvir Crncevic <elvircrn@gmail.com>
Signed-off-by: Tyler Michael Smith <tlrmchlsmth@gmail.com>
Signed-off-by: Elvir Crnčević <elvircrn@gmail.com>
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Co-authored-by: Tyler Michael Smith <tlrmchlsmth@gmail.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
2026-08-25 10:21:00 -04:00 |
|
|
Harry MellorandClaude Opus 5
|
b2dd9ce73d
|
[Docs] Fix docstring continuation indentation in routed_experts.py (#53604)
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
|
2026-08-25 02:27:16 +00:00 |
|
|
kiroxuandOpenAI Codex
|
07ef21bc69
|
[Kernel][Perf] Enable fused QK-norm + partial MRoPE + gate for Qwen3.6 (#52676)
Signed-off-by: kiroxu <148877251+BabyDrangoner@users.noreply.github.com>
Co-authored-by: OpenAI Codex <codex@openai.com>
|
2026-08-24 18:28:49 -07:00 |
|
|
Giancarlo Delfin
|
d154d90d6c
|
[Model Runner V2] batch-sharded sample (#50465)
Signed-off-by: Giancarlo Delfin <gdelfin@inferact.ai>
|
2026-08-24 15:49:57 -07:00 |
|
|
Wentao Ye
|
f59bb0bd8c
|
[Refactor] Refactor batch invariance folder (#53619)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
|
2026-08-24 14:31:36 -07:00 |
|
|
Lio Einaudiandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
7797b6022c
|
[Kernel][Perf] Per-architecture tuned configs for batch-invariant persistent matmul (~3x decode kernels on RTX 4090D/H20) (#53247)
Signed-off-by: LioEinaudi <zhao3024667639@gmail.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
2026-08-24 11:28:51 -04:00 |
|
|
Wentao Ye
|
29c9af5211
|
[Kimi K3 Refactor] Add UnfinalizedMoEOutput proto following up for #53152 (#53310)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
|
2026-08-24 11:21:11 -04:00 |
|
|
cjackal
|
22099afc64
|
[Bugfix][DCP] Handle sparse MLA metadata after DCP Manager refactor (#52377)
Signed-off-by: cjackal <44624812+cjackal@users.noreply.github.com>
|
2026-08-24 15:02:48 +00:00 |
|
|
Taneem Ibrahim
|
79bb395eea
|
[Pooling] Improve BGE-M3 sync pooling throughput by up to 3.13% (#53464)
Signed-off-by: Taneem Ibrahim <taneem.ibrahim@gmail.com>
|
2026-08-24 08:33:32 -04:00 |
|
|
|
a4d70bef37
|
[Model Runner V2] Reserve CUDA graph memory (#53306)
Signed-off-by: Anh Tran <anh.tran.3889@gmail.com>
Signed-off-by: Nick Hill <nickhill123@gmail.com>
Co-authored-by: Anh Tran <anh.tran.3889@gmail.com>
Co-authored-by: anhtra3889 <anhtra@nvidia.com>
|
2026-08-23 23:11:19 -07:00 |
|
|
ℍ𝕠𝕝𝕝𝕠𝕨 𝕄𝕒𝕟andJee Jee Li
|
702e1d7186
|
[LoRA] feat: Support LoRA for DeepSeek V4 (#53361)
Signed-off-by: Hollow Man <hollowman@opensuse.org>
Co-authored-by: Jee Jee Li <pandaleefree@gmail.com>
|
2026-08-24 03:57:57 +00:00 |
|
|
|
da329cc303
|
[Bugfix] Fix speculative decoding for short_conv (LFM2) models (#50272)
Signed-off-by: jens.luecke <jens.luecke@liquid.ai>
Co-authored-by: Tonoken3 <sakamakismile@gmail.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
2026-08-22 13:34:22 +00:00 |
|
|
ℍ𝕠𝕝𝕝𝕠𝕨 𝕄𝕒𝕟
|
b2db227a7c
|
[Bugfix][R3] Unwrap UniformTypeKVCacheSpecs when selecting the routed-experts KV group (#53240)
Signed-off-by: Hollow Man <hollowman@opensuse.org>
|
2026-08-21 19:13:53 -07:00 |
|
|
stefankoncarevic
|
7f4a1b7e24
|
[Bugfix][ROCM] Fix the MXFP8 block scale exponent (#53110)
Signed-off-by: Stefan Koncarevic <Stefan.Koncarevic@amd.com>
|
2026-08-22 01:26:19 +00:00 |
|
|
Lucas Wilkinson
|
8bdc70ec7b
|
[6/N][KV-Cache Layout Refactor] Standardize KV cache layout (#51718)
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
|
2026-08-21 17:47:49 -07:00 |
|
|
vllm-agentandvllm-ci-failure-analyzer
|
ba53da60bb
|
Revert "[Bugfix][MoE] Tune FlashInfer experts to scheduler token limit" (#52989) (#53186)
Co-authored-by: vllm-ci-failure-analyzer <kevin@inferact.ai>
|
2026-08-21 12:39:28 -06:00 |
|
|
|
7a2fdbaac4
|
[K3 Perf] Fuse MXFP4 top-k finalization into latent-tail, ~5% E2E latency reduction (#53152)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
Signed-off-by: Yongye Zhu <zyy1102000@gmail.com>
Co-authored-by: Yongye Zhu <zyy1102000@gmail.com>
Co-authored-by: OpenAI Codex <codex@openai.com>
|
2026-08-21 13:29:59 -04:00 |
|
|
Fangzhou Ai
|
fe76112ff2
|
[ROCm][Perf] Optimize DeepSeek V4 C4A top-k with AITER (#52882)
|
2026-08-21 10:00:39 -07:00 |
|
|
ErenAta16
|
fbb17e780b
|
[Bugfix] Fix six quantization exception messages split across positional args (#50479)
Signed-off-by: ErenAta16 <erena6466@gmail.com>
|
2026-08-21 09:06:56 -07:00 |
|
|
Luke Alonsoandmgoin
|
2740c817ff
|
[Kernel] Add b12x FP4 MoE backend (#52018)
Signed-off-by: Luke Alonso <lalonso@gmail.com>
Signed-off-by: mgoin <mgoin64@gmail.com>
Signed-off-by: Misha Goin <mgoin64@gmail.com>
Co-authored-by: mgoin <mgoin64@gmail.com>
|
2026-08-21 08:04:36 -07:00 |
|
|
Yan Ma
|
6feafb8b7d
|
[XPU] follow cuda path for mrope on XPU (#53201)
Signed-off-by: Yan Ma <yan.ma@intel.com>
|
2026-08-21 10:13:27 +00:00 |
|
|
Thanaji Rao ThakkalapelliandKunshang Ji
|
9fd750f00d
|
[XPU][INC] Add int4 w4a8 (dynamic int8 activation) backend for INC linear layers (#50501)
Signed-off-by: Thanaji Rao Thakkalapelli <thanaji.rao.thakkalapelli@intel.com>
Co-authored-by: Kunshang Ji <kunshang.ji@intel.com>
|
2026-08-21 14:38:59 +08:00 |
|
|
Wei ZhaoandOpenAI Codex
|
f8e0602713
|
Support kimi k3 nvfp4 checkpoint (#53132)
Signed-off-by: wzhao18 <wzhao18.sz@gmail.com>
Co-authored-by: OpenAI Codex <codex@openai.com>
|
2026-08-21 05:59:44 +00:00 |
|
|
Zihan Zhang
|
b389ac2946
|
[Spec Decode] DFlash2: local convolution + candidate selector (#52816)
|
2026-08-20 22:27:22 -07:00 |
|
|
vhagor
|
ba07e4a48f
|
[Bugfix] Fix batch-invariant fp32 matmul OOR on SM89 for N=1 (#52960)
Signed-off-by: Juqi Li <2223621784@qq.com>
|
2026-08-21 11:46:55 +08:00 |
|
|
Misha GoinandOpenAI Codex
|
bfb6c13499
|
[Bugfix][MoE] Tune FlashInfer experts to scheduler token limit (#52989)
Signed-off-by: mgoin <mgoin64@gmail.com>
Co-authored-by: OpenAI Codex <codex@openai.com>
|
2026-08-20 16:26:24 -04:00 |
|
|
pmanczak
|
44cf3f0466
|
[XPU][Tests] Make tests device-agnostic (#51968)
Signed-off-by: pmanczak <pawel.manczak@intel.com>
|
2026-08-20 02:25:41 -07:00 |
|
|
Anton AandRoger Wang
|
14617c2b6c
|
[Bugfix] DeepEP-V2: expert_tokens_meta must be None on the decode/cudagraph path (empty recv_expert_num_tokens) (#52632)
Signed-off-by: Anton Alexander <dmvevents@gmail.com>
Signed-off-by: Roger Wang <hey@rogerw.io>
Co-authored-by: Roger Wang <hey@rogerw.io>
|
2026-08-20 01:11:44 -07:00 |
|
|
Shanshan Shen
|
d626108b18
|
[ROCm][Perf] Fuse DeepSeek-V4 mHC post/pre and RMSNorm with AITER (#52737)
Signed-off-by: shen-shanshan <467638484@qq.com>
Signed-off-by: Shanshan Shen <87969357+shen-shanshan@users.noreply.github.com>
|
2026-08-20 04:51:04 +00:00 |
|
|
Summer YangandWoosuk Kwon
|
58e5ee0158
|
[refactor] consolidate cp attn ops (#52839)
Signed-off-by: Summer Yang <girasoleyang@gmail.com>
Co-authored-by: Woosuk Kwon <woosuk@inferact.ai>
|
2026-08-19 18:04:11 -07:00 |
|
|
Sage Moore
|
5bf0dbd6fd
|
[Bugfix] vLLM crashes at startup when DeepEP v2 is used with --enforce-eager wiht TRTLLM Bf16 (#51824)
|
2026-08-19 20:18:47 -04:00 |
|
|
Roger Wang
|
755492e37d
|
Revert "[Kernel] Gemma-4 FA4 FP8 Kernel" (#52987)
|
2026-08-19 10:48:46 -07:00 |
|
|
|
54dd98be28
|
[CT] Support Humming for WNA16 MoE (#48918)
Signed-off-by: yiliu30 <yi4.liu@intel.com>
Co-authored-by: OpenAI Codex <codex@openai.com>
Co-authored-by: OpenAI Codex <noreply@openai.com>
Co-authored-by: Misha Goin <mgoin64@gmail.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
2026-08-19 12:31:32 -04:00 |
|
|
xuebwang-amd
|
c676232313
|
[Bugfix][Quantization] Fix OCP MX MoE emulation silently skipping mxfp6 activation QDQ (#52704)
Signed-off-by: xuebwang-amd <xuebwang@amd.com>
|
2026-08-19 11:29:29 -05:00 |
|