 Xinyu Chen
|
3b1d440ede
|
CustomOp: grouped topk (#29575)
Signed-off-by: Xinyu Chen <xinyu1.chen@intel.com>
|
2025-12-17 17:43:00 +08:00 |
|
|
Yan Ma
|
4f735babb7
|
[XPU] fix broken fp8 online quantization for XPU platform (#30831)
Signed-off-by: Yan Ma <yan.ma@intel.com>
|
2025-12-17 00:28:13 -08:00 |
|
|
Li, Jiangandgemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
|
0cd5353644
|
[Bugfix][CPU] Fix CPU backend ROPE dispatch for VL models (#30829)
Signed-off-by: jiang1.li <jiang1.li@intel.com>
Signed-off-by: Li, Jiang <bigpyj64@gmail.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
|
2025-12-16 23:25:12 -08:00 |
|
|
Jinzhen LinandMichael Goin
|
ce96857fdd
|
[Kernel][Quantization][MoE] add marlin kernel support for turing (sm75) (#29901)
Signed-off-by: Jinzhen Lin <jinzhen.ljz@antgroup.com>
Co-authored-by: Michael Goin <mgoin64@gmail.com>
|
2025-12-16 14:35:28 -08:00 |
|
|
jiahanc
|
254a7f8fd6
|
[Perf] Do FP4 quant before All gather on flashinfer trtllmgen MOE (#30014)
Signed-off-by: jiahanc <173873397+jiahanc@users.noreply.github.com>
|
2025-12-16 13:01:48 -08:00 |
|
|
Michael Goin
|
10ee1c64cf
|
[CI] Generalize gsm8k test args and add Qwen3-Next MTP B200 test (#30723)
Signed-off-by: mgoin <mgoin64@gmail.com>
|
2025-12-16 14:28:34 -05:00 |
|
|
Ming Yang
|
ce12b407f2
|
[TRTLLM] Remove the MoE GEMM weight name change (#30713)
Signed-off-by: Ming Yang <minos.future@gmail.com>
|
2025-12-16 11:01:38 -05:00 |
|
|
Wentao Ye
|
59bd5f6a71
|
[Feat] Enable eplb with default all2all backend (#30559)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
|
2025-12-16 10:33:52 -05:00 |
|
|
|
3bd9c49158
|
[CustomOp] Extract ApplyRotaryEmb as CustomOp and unify the dispatch logic (#29873)
Signed-off-by: shen-shanshan <467638484@qq.com>
Co-authored-by: gcanlin <canlinguosdu@gmail.com>
Co-authored-by: TJian <tunjian.tan@embeddedllm.com>
|
2025-12-15 19:08:16 -08:00 |
|
|
Matthew Bonanniandgemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
|
60dbf7d8f1
|
Update batch invariant to use attention config (#30704)
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
|
2025-12-15 15:24:16 -05:00 |
|
|
Robert ShawandRobert Shaw
|
d0502b4928
|
[MoE][Refactor 1/N] Separate Online Quantization (#30627)
Signed-off-by: Robert Shaw <robshaw@redhat.com>
Co-authored-by: Robert Shaw <robshaw@redhat.com>
|
2025-12-15 06:54:53 -08:00 |
|
|
|
3778673ea8
|
[Feat] Refactor for parallel_config in FusedMoEModularKernel (#30282)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
Signed-off-by: Robert Shaw <robshaw@redhat.com>
Co-authored-by: Robert Shaw <robshaw@redhat.com>
Co-authored-by: Robert Shaw <114415538+robertgshaw2-redhat@users.noreply.github.com>
|
2025-12-15 04:21:36 +00:00 |
|
|
tjp_zju
|
6ecc1e411b
|
[Bugfix] fix _get_quant_method of FusedMoE for deepseekV3.2 on non-NV… (#30057)
Signed-off-by: tjp_zju <tanjianpingzju1990@gmail.com>
|
2025-12-14 02:20:51 -08:00 |
|
|
Shengliang XuandPavani Majety
|
0bb0bae436
|
Nvidia ModelOpt workaround for issue 28072 (#30164)
Signed-off-by: Shengliang Xu <shengliangx@nvidia.com>
Co-authored-by: Pavani Majety <pmajety@nvidia.com>
|
2025-12-14 18:18:31 +08:00 |
|
|
Didier DurandandWentao Ye
|
1a55cfafcb
|
[Doc]: fixing typos in various files (#30540)
Signed-off-by: Didier Durand <durand.didier@gmail.com>
Signed-off-by: Didier Durand <2927957+didier-durand@users.noreply.github.com>
Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
|
2025-12-14 02:14:37 -08:00 |
|
|
Wentao Yeandgemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
|
6e78ed6ba7
|
[Logs] Optimize startup logs 4 (#29903)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
Signed-off-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
|
2025-12-13 16:12:53 -05:00 |
|
|
Tsukasa OI
|
fdc135d768
|
[Misc][Quantization] Clarify the intent of GGUF FusedMoE weight materialization (#30310)
Signed-off-by: Tsukasa OI <floss_llm@irq.a4lg.com>
|
2025-12-13 13:55:14 +08:00 |
|
|
Roberto L. Castroandyoukaichao
|
4fa7ce46f3
|
[Feature] Add SM103 (Blackwell Ultra) Support to vLLM (#30484)
Signed-off-by: LopezCastroRoberto <robertol.c510@gmail.com>
Signed-off-by: Roberto L. Castro <38211239+LopezCastroRoberto@users.noreply.github.com>
Co-authored-by: youkaichao <youkaichao@gmail.com>
|
2025-12-12 19:34:23 -08:00 |
|
|
rasmithandRandall Smith
|
08f8a5627e
|
[CI/Build][Kernel][BugFix][AMD] Fix per_token_group_quant_fp8 to use correct fp8 min/max values and update atol/rtol in test_quantfp8_group_functionality (#30292)
Signed-off-by: Randall Smith <ransmith@amd.com>
Co-authored-by: Randall Smith <ransmith@amd.com>
|
2025-12-12 18:41:56 -05:00 |
|
|
|
13618626df
|
[MoE-FP8-modelopt] Add FlashInfer alignment padding for intermediate dimensions (#29748)
Signed-off-by: Daniel Afrimi <dafrimi@pool0-00589.cm.cluster>
Signed-off-by: dafrimi <dafrimi@nvidia.com>
Co-authored-by: Daniel Afrimi <dafrimi@pool0-00589.cm.cluster>
Co-authored-by: Tyler Michael Smith <tyler@neuralmagic.com>
|
2025-12-12 20:42:32 +00:00 |
|
|
Xin Yang
|
1f19d8f899
|
[Perf] Set split_k to 1 for triton_kernels (#30528)
Signed-off-by: Xin Yang <xyangx@amazon.com>
|
2025-12-12 14:07:57 -05:00 |
|
|
shivampr
|
cd7740ac5c
|
[ROCm] Enable Triton ScaledMM fallback + kernel selection fix (#26668)
Signed-off-by: Shivam <shivampr.dev@gmail.com>
Signed-off-by: Shivam <shivamprasad91@gmail.com>
|
2025-12-12 13:28:20 -05:00 |
|
|
|
dc13c99eed
|
fix(gguf): Disable bfloat16 for GGUF on blackwell device (#30408)
Signed-off-by: Christina <truffle@gmail.com>
Signed-off-by: Isotr0py <2037008807@qq.com>
Signed-off-by: Christina Norman <christina@example.com>
Co-authored-by: Isotr0py <isotr0py@users.noreply.github.com>
Co-authored-by: Isotr0py <2037008807@qq.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
|
2025-12-12 23:10:12 +08:00 |
|
|
Lucas Wilkinson
|
3e41992fec
|
[Attention] Use sparse prefill kernel for fp8 kv-cache in DeepSeek-v3.2 (#27532)
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
|
2025-12-12 05:57:47 -08:00 |
|
|
Bhanu Prakash Voutharoja
|
6a6fc41c79
|
gptq marlin quantization support for fused moe with lora (#30254)
Signed-off-by: Bhanu068 <voutharoja.bhanu06@gmail.com>
|
2025-12-12 02:27:22 +00:00 |
|
|
jiahanc
|
0ab23c2b2b
|
[fix] fix SM check for Flashinfer TRTLLM MOE (#30314)
Signed-off-by: jiahanc <173873397+jiahanc@users.noreply.github.com>
|
2025-12-12 01:00:58 +00:00 |
|
|
Andrew BriandandAndrew Briand
|
a00d88973d
|
[EPLB] Support EPLB w/ NVFP4 (#29804)
Signed-off-by: Andrew Briand <abriand@nvidia.com>
Co-authored-by: Andrew Briand <abriand@nvidia.com>
|
2025-12-11 22:59:40 +00:00 |
|
|
|
c817b14151
|
[Perf] Optimize deepgemm experts initialization, 3.9% TTFT improvement (#30494)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
Co-authored-by: li-jinpeng <3332126450@qq.com>
Co-authored-by: youkaichao <youkaichao@gmail.com>
|
2025-12-11 17:28:34 -05:00 |
|
|
Harry Mellor
|
cf3eacfe58
|
Standardise get_rope to use rope_parameters["partial_rotary_factor"], not rotary_dim (#30389)
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
|
2025-12-11 20:45:23 +00:00 |
|
|
汪志鹏
|
0e71eaa644
|
[Feature] AWQ marlin quantization support for fused moe with lora (#30442)
Signed-off-by: princepride <wangzhipeng628@gmail.com>
|
2025-12-11 18:03:32 +00:00 |
|
|
|
d1e1fb4363
|
[Bugfix] Fix grouped_topk pytorch impl when num_experts can't be grouped properly (#29439)
Signed-off-by: Divakar Verma <divakar.verma@amd.com>
Co-authored-by: Gregory Shtrasberg <156009573+gshtras@users.noreply.github.com>
Co-authored-by: TJian <tunjian.tan@embeddedllm.com>
|
2025-12-10 19:47:18 -08:00 |
|
|
Wilson WuandWentao Ye
|
3bdd426636
|
Fix typos in comments across multiple files (#30345)
Signed-off-by: Wilson Wu <iwilsonwu@gmail.com>
Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
|
2025-12-09 20:05:28 -08:00 |
|
|
ElizaWszola
|
2e7035dd8c
|
[Bugfix] Fix fp8 DeepGemm compilation issues (#30336)
|
2025-12-09 20:17:25 -05:00 |
|
|
|
3c680f4a17
|
[Rocm][torch.compile] Adding layernorm + fp8 block quant and silu + fp8 block quant for Aiter (#25693)
Signed-off-by: charlifu <charlifu@amd.com>
Signed-off-by: Micah Williamson <micah.williamson@amd.com>
Signed-off-by: Charlie Fu <Charlie.Fu@amd.com>
Co-authored-by: Micah Williamson <micah.williamson@amd.com>
Co-authored-by: wuhuikx <hattie.wu@amd.com>
Co-authored-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
Co-authored-by: Gregory Shtrasberg <156009573+gshtras@users.noreply.github.com>
|
2025-12-09 22:39:26 +00:00 |
|
|
Kyle Sayers
|
fccd532587
|
[Quantization] FP8 Weight Reloading for Quantized RL Rollout (#28480)
Signed-off-by: Kyle Sayers <kylesayrs@gmail.com>
|
2025-12-09 13:54:32 -08:00 |
|
|
|
00e5cbb967
|
[MoE][Refactor] Remove most arguments to FusedMoEMethodBase.apply (#29066)
Signed-off-by: Bill Nell <bnell@redhat.com>
Signed-off-by: Tyler Michael Smith <tlrmchlsmth@gmail.com>
Co-authored-by: Robert Shaw <114415538+robertgshaw2-redhat@users.noreply.github.com>
Co-authored-by: Lucas Wilkinson <LucasWilkinson@users.noreply.github.com>
Co-authored-by: Tyler Michael Smith <tlrmchlsmth@gmail.com>
|
2025-12-09 13:48:25 -08:00 |
|
|
quanliuandWentao Ye
|
5dcd593baf
|
[Feature] Batch-Invariant Support for FA2 and LoRA (#30018)
Signed-off-by: quanliu <18646313696@163.com>
Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
|
2025-12-09 10:01:38 -05:00 |
|
|
vllmellm
|
ee14644ba9
|
[ROCm] Aiter Quant Kernels (#25552)
Signed-off-by: vllmellm <vllm.ellm@embeddedllm.com>
|
2025-12-09 14:27:37 +00:00 |
|
|
Dongjie Zou
|
1166c31cc7
|
[Bugfix]: Fix glm46 awq marlin moe wna16 compatibility (#30210)
Signed-off-by: baonudesifeizhai <baonudesifeizhai@gmail.com>
|
2025-12-09 12:20:21 +00:00 |
|
|
Tsukasa OIandIsotr0py
|
58d5b3f514
|
[Model][Quantization] Restore MoE + GGUF models support (incl. Qwen3 MoE) by allowing Sideload Parameters (#30116)
Signed-off-by: Tsukasa OI <floss_llm@irq.a4lg.com>
Co-authored-by: Isotr0py <mozf@mail2.sysu.edu.cn>
|
2025-12-09 05:30:05 +00:00 |
|
|
Michael Goinandgemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
|
03b91f7262
|
[Bugfix] Fix compressed-tensors models failing to load with transformers backend (#30287)
Signed-off-by: mgoin <mgoin64@gmail.com>
Signed-off-by: Michael Goin <mgoin64@gmail.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
|
2025-12-08 20:44:28 -08:00 |
|
|
czhu-cohere
|
f6227c22ab
|
[Kernel]Support W4A8 Grouped GEMM on Hopper (#29691)
Signed-off-by: czhu-cohere <conway.zhu@cohere.com>
|
2025-12-08 19:29:06 -08:00 |
|
|
Zhewen Li
|
ae339b1a67
|
[Bugfix] Fix DeepGEMM after #29546 (#30267)
Signed-off-by: zhewenli <zhewenli@meta.com>
Signed-off-by: Zhewen Li <zhewenli@meta.com>
|
2025-12-09 01:05:27 +00:00 |
|
|
Wentao Ye
|
d9417096d1
|
[Feature] Batch invariant: Enable TRITON_MLA without prefix-caching (#29125)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
|
2025-12-08 19:31:57 -05:00 |
|
|
Ming Yang
|
9d6235ca9a
|
[moe] Allow disabling DP chunking (#29936)
Signed-off-by: Ming Yang <minos.future@gmail.com>
|
2025-12-09 00:29:36 +00:00 |
|
|
roikoren755
|
ae0f69b16a
|
Add SpecDec support to selective_state_update (#29488)
Signed-off-by: Roi Koren <roik@nvidia.com>
|
2025-12-08 16:45:18 -05:00 |
|
|
Vasiliy Kuznetsov
|
0d402d2600
|
online fp8 quant with streaming weight post-processing (#29196)
Signed-off-by: vasiliy <vasiliy@fb.com>
|
2025-12-08 20:15:10 +00:00 |
|
|
Zhiwei
|
c6df05ebb4
|
[ROCm] [Fused Moe EP] Use binary expert mask for aiter fused moe kernel (#29773)
Signed-off-by: ZhiweiYan-96 <zhiwei.yan@amd.com>
|
2025-12-08 05:23:46 +00:00 |
|
|
ElizaWszolaandyewentao256
|
af0444bf40
|
[Performance] Fused blockwise quant RMS norm (#27883)
Signed-off-by: ElizaWszola <ewszola@redhat.com>
Signed-off-by: yewentao256 <zhyanwentao@126.com>
Co-authored-by: yewentao256 <zhyanwentao@126.com>
|
2025-12-07 16:38:04 +00:00 |
|
|
Wentao Ye
|
541a2ef892
|
[Perf] Deepgemm fused layout kernel for activations, 4.3% throughput improvement, 10.7% TTFT improvement. (#29546)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
|
2025-12-07 20:31:14 +08:00 |
|