kanshan/TensorRT-LLMs
1
0
Fork 0
mirror of https://github.com/NVIDIA/TensorRT-LLM.git synced 2026-02-21 18:25:20 +08:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity
4,440 Commits 29 Branches 69 Tags 2 GiB
bdf6953ddc
Commit Graph

1086 Commits

Author SHA1 Message Date
Izzy Putterman
bdf6953ddc
[None][feat] Eagle: MLA Based Eagle (#9677) 
Signed-off-by: Izzy Putterman <iputterman@nvidia.com>
 2026-01-02 13:45:07 -05:00
Balaram Buddharaju
4a1b742aa0
[TRTLLM-9467][fix] Fix PP+CP combination with helix parallelism (#10312) 
Signed-off-by: Balaram Buddharaju <169953907+brb-nv@users.noreply.github.com>
 2026-01-01 13:42:53 -05:00
Lucas Liebenwein
1bbe71b3ed
[#10244][feat] AutoDeploy: separate prefill/decode in flashinfer (#10252) 
Signed-off-by: Lucas Liebenwein <11156568+lucaslie@users.noreply.github.com>
 2025-12-31 17:01:24 -05:00
Simeng Liu
84d107b2f0
[https://nvbugs/5717993][fix] Add execution_stream across PyExecutor, KVCacheManager, PeftCacheManager to ensure proper CUDA stream synchronization between KV cache transfer operations and model forward kernels. (#10060) 
Signed-off-by: SimengLiu-nv <simengl@nvidia.com>
 2025-12-31 09:22:54 -08:00
tcherckez-nvidia
464847c6be
[#9717][chore] Standardize MoE weights interface (#10295) 
Signed-off-by: Tal Cherckez <127761168+tcherckez-nvidia@users.noreply.github.com>
 2025-12-31 07:37:18 -05:00
Necofish
73870ae4ad
[None][feat] support Qwen3-VL dense model in pytorch backend (#9060) 
Signed-off-by: Nekofish-L <liuxiangyang@mail.ustc.edu.cn>
 2025-12-31 17:54:26 +09:00
Neta Zmora
966231d29c
[#9626][feat] Add an auto-deploy transform for using cutlass FP4 MoE kernels (#10304) 
Add a transform to relace torch.ops.auto_deploy.torch_quant_nvfp4_moe
with the optimized torch.ops.auto_deploy.trtllm_quant_nvfp4_moe_fused.

Currently generates the wrong results when the number of rows in MoE FC1 weights is not divisible by 128,
so torch.ops.auto_deploy.trtllm_quant_nvfp4_moe_fused is not set as the default FP4 MoE implementation (i.e. the transform is disabled).

Signed-off-by: Neta Zmora <96238833+nzmora-nvidia@users.noreply.github.com>
 2025-12-29 23:18:15 +02:00
JunyiXu-nv
55bc6a5ff8
[https://nvbugs/5753250][fix] Fix undefined local variable in responses utils (#10154) 
Signed-off-by: Junyi Xu <219237550+JunyiXu-nv@users.noreply.github.com>
Signed-off-by: JunyiXu-nv <219237550+JunyiXu-nv@users.noreply.github.com>
 2025-12-28 06:59:32 +08:00
shivghai
ee07a7c55e
[None][fix] [Gemma3] Fix RoPE for local attention for Gemma3 (#9961) 
Signed-off-by: Shiv Ghai <8965168+shivghai@users.noreply.github.com>
 2025-12-27 11:50:59 -08:00
Guoming Zhang
93ac0bc1dc
[TRTLLM-10126][feat] Increase topk upper limit to 22 for NVLinkOneSid… (#10229) 
Signed-off-by: nv-guomingz <137257613+nv-guomingz@users.noreply.github.com>
 2025-12-27 22:48:10 +08:00
Pengyun Lin
684b37df02
[https://nvbugs/5747938][fix] Use local tokenizer (#10230) 
Signed-off-by: Pengyun Lin <81065165+LinPoly@users.noreply.github.com>
 2025-12-26 22:08:10 +08:00
Pengyun Lin
c5b0f9e436
[https://nvbugs/5633700][fix] Cache tiktoken vocab for gpt-oss (#10219) 
Signed-off-by: Pengyun Lin <81065165+LinPoly@users.noreply.github.com>
 2025-12-26 18:39:03 +08:00
Neta Zmora
f3f02315df
[None][chore]: small refactoring to auto-deploy MoE operator (#10300) 
Signed-off-by: Neta Zmora <96238833+nzmora-nvidia@users.noreply.github.com>
 2025-12-25 12:27:11 -05:00
Ziyi Xiong
d8b5aeb061
[https://nvbugs/5652062][fix] Rewind kv_cache and reset draft tokens (#10160) 
Signed-off-by: ziyixiong-nv <219238287+ziyixiong-nv@users.noreply.github.com>
 2025-12-25 09:13:51 -05:00
ZhichenJiang
46e4af5688
[TRTLLM-9831][perf] Enable 2CTA with autotune for CuteDSL MoE and Grouped GEMM optimizations (#10201) 
Signed-off-by: zhichen jiang <zhichenj@NVIDIA.com>
Signed-off-by: Enwei Zhu <21126786+syuoni@users.noreply.github.com>
Co-authored-by: Enwei Zhu <21126786+syuoni@users.noreply.github.com>
 2025-12-25 09:04:20 -05:00
Gabriel Wu
1d01214ff0
[None][feat] Drop non-deepgemm fp8 block scale gemm (#10256) 
Signed-off-by: Zihua Wu <13583761+lucifer1004@users.noreply.github.com>
 2025-12-25 14:52:52 +08:00
Venky
c059e6caa1
[TRTC-121] [feat] Add recipe selector UI to complement the recipe database (#10125) 
Signed-off-by: Venky Ganesh <23023424+venkywonka@users.noreply.github.com>
 2025-12-24 23:56:54 -05:00
Neta Zmora
c4b36d31ff
[#10137][feat] AutoDeploy FP8 MoE refactor (#10138) 
The trtllm (cutlass) fp8 moe operator performs W3+W1 fusion (concat) during inference and we want to move this fusion to the model optimization time.

The Cutlass MoE kernel is used thru a trtllm torch operator.
Its implementation uses two FC operations (fc1 and fc2) while the canonical MoE API defines three GEMM operations and their associated weights (W1, W2, W3) so when we switch from the torch.moe op to the trtllm.moe op we also change terminology from w1, w2, w3 to fc1, fc2.

Signed-off-by: Neta Zmora <96238833+nzmora-nvidia@users.noreply.github.com>
 2025-12-24 18:58:10 +02:00
shuyixiong
f4f0fe85e9
[TRTLLM-9737][chore] Add rl perf reproduce script and enhance the robustness of Ray tests (#9939) 
Signed-off-by: Shuyi Xiong <219646547+shuyixiong@users.noreply.github.com>
 2025-12-24 15:27:01 +08:00
Fanrong Li
156f6453dc
[TRTLLM-9798][feat] Change to use new DeepGEMM MQA sm100 kernel for MTP-3 (#10226) 
Signed-off-by: Fanrong Li <23290157+lfr-0531@users.noreply.github.com>
 2025-12-24 14:39:12 +08:00
Balaram Buddharaju
8c1cfc872b
[TRTLLM-9493][feat] Custom AllToAll for helix parallelism (#9986) 
Signed-off-by: Balaram Buddharaju <169953907+brb-nv@users.noreply.github.com>
 2025-12-23 18:14:30 -08:00
Shiyu Li
3ddc9d2b48
[https://nvbugs/5729697][fix] MNNVL Allreduce: use CUDA runtime instead of Macro to get SM version. (#10062) 
Signed-off-by: Shiyu Li <shili@nvidia.com>
 2025-12-23 16:07:07 +08:00
Bo Li
cc1323be24
[None][fix] Fix the bug for top_k=10 in NVLinkOneSided AlltoAll. (#10197) 
Signed-off-by: Bo Li <22713281+bobboli@users.noreply.github.com>
 2025-12-23 02:13:37 -05:00
Harshini Komali
d691371eaf
[TRTLLM-9091] [feat] Replace GenAI-Perf with AIPerf (#9310) 
Signed-off-by: lkomali <lkomali@nvidia.com>
Signed-off-by: Harshini Komali <157742537+lkomali@users.noreply.github.com>
Co-authored-by: Kaiyu Xie <26294424+kaiyux@users.noreply.github.com>
 2025-12-23 13:25:55 +08:00
Yuxian Qiu
696f754ef4
[None][fix] avoid implicit cudaStreamSynchronize in sample_async. (#10120) 
Signed-off-by: Yuxian Qiu <142763828+yuxianq@users.noreply.github.com>
 2025-12-23 10:15:40 +08:00
Fanrong Li
0d2500c631
[TRTLLM-9677][feat] Support DeepSeek-V3.2 tool parser (#10126) 
Signed-off-by: Fanrong Li <23290157+lfr-0531@users.noreply.github.com>
 2025-12-23 08:46:47 +08:00
tcherckez-nvidia
12e1cb8d7e
[#9717][chore] Refactor MoE code to use enums (#9910) 
Signed-off-by: Tal Cherckez <127761168+tcherckez-nvidia@users.noreply.github.com>
 2025-12-22 15:14:56 -05:00
JunyiXu-nv
aaa87abf41
[TRTLLM-7906][feat] Support multiple post process for Responses API (#9908) 
Signed-off-by: Junyi Xu <219237550+JunyiXu-nv@users.noreply.github.com>
 2025-12-22 11:33:34 -05:00
Pengyun Lin
0f308e95f9
[None][chore] Remove logprobs constraint on trtllm-serve pytorch backend (#9911) 
Signed-off-by: Pengyun Lin <81065165+LinPoly@users.noreply.github.com>
 2025-12-22 21:37:22 +08:00
William Zhang
a6a88985cf
[TRTLLM-9409][feat] Pass MRoPE tensors for EPD disagg (#9758) 
* Why?

Certain VLMs like the Qwen family need more than just the multimodal
embeddings in the language model, and need MRoPE position IDs and
deltas. Prior to this commit, only the embeddings could be communicated
from the encoder worker to the prefill worker.

* What?

This commit extends the `DisaggregatedParams` to include the MRoPE
information. It also adjusts several pieces of code required to
communicate that between E, P and D workers.

Closes TRTLLM-9409.

Signed-off-by: William Zhang <133824995+2ez4bz@users.noreply.github.com>
 2025-12-22 06:32:49 -05:00
Bo Li
472fe497dc
[None][chore] NVLinkOneSided AlltoAll Support zero local_num_tokens. (#9822) 
Signed-off-by: Bo Li <22713281+bobboli@users.noreply.github.com>
 2025-12-22 05:57:12 -05:00
Yan Chunwei
ea6cd76c55
[None][refactor] simplify get_stats and get_kvcache_events with rpc (#9980) 
Signed-off-by: Yan Chunwei <328693+Superjomn@users.noreply.github.com>
Signed-off-by: Superjomn <328693+Superjomn@users.noreply.github.com>
 2025-12-22 18:23:43 +08:00
shuyixiong
9e9523c3cc
[https://nvbugs/5762016][chore] Skip a ray test (#10194) 
Signed-off-by: Shuyi Xiong <219646547+shuyixiong@users.noreply.github.com>
 2025-12-22 17:06:19 +08:00
Balaram Buddharaju
5266475014
[None][feat] Cudagraph updates for helix parallelism (#10141) 
Signed-off-by: Balaram Buddharaju <169953907+brb-nv@users.noreply.github.com>
 2025-12-21 15:21:52 -05:00
xxi
5ae154022a
[TRTLLM-9872][fix] clear the failed test at CI when enalbe_configurab… (#10067) 
Signed-off-by: xxi <xxi@nvidia.com>
 2025-12-21 08:14:50 -05:00
Bo Li
77e37d9dd0
[https://nvbugs/5753250][infra] Further waive all tests in _test_openai_responses.py (#10176) 
Signed-off-by: Bo Li <22713281+bobboli@users.noreply.github.com>
 2025-12-20 10:25:14 -05:00
Enwei Zhu
2ce785f39a
[https://nvbugs/5643631][fix] Fix hostfunc seg fault (#10028) 
Signed-off-by: Enwei Zhu <21126786+syuoni@users.noreply.github.com>
 2025-12-20 07:58:43 -05:00
Anish Shanbhag
7c82605327
[None][fix] enable KV cache reuse for config database (#10094) 2025-12-19 15:16:56 -08:00
longcheng-nv
b882393d69
[https://nvbugs/5720357][fix] Fix indice offset overflow in custom Top-K kernel and corresponding UT case (#10027) 
Signed-off-by: longcheng-nv <243710427+longcheng-nv@users.noreply.github.com>
Co-authored-by: Chang Liu (Enterprise Products) <9713593+chang-l@users.noreply.github.com>
 2025-12-19 14:58:01 -05:00
Venky
dfa11d810e
[TRTC-102][docs] --extra_llm_api_options->--config in docs/examples/tests (#10005) 2025-12-19 13:48:43 -05:00
Emma Qiao
dd8ce68c94
[None][infra] Update waive and waive failed tests for main branch on 12/19 (#10151) 
Signed-off-by: qqiao <qqiao@nvidia.com>
 2025-12-19 01:20:42 -08:00
Pengyun Lin
ac03915dc3
[TRTLLM-9604][feat] DS R1 & V3.1 tool parser (#10010) 
Signed-off-by: Pengyun Lin <81065165+LinPoly@users.noreply.github.com>
 2025-12-19 17:20:03 +08:00
Chang Liu
31bc14b350
[TRTLLM-9654][feat] Support DeepSeek-V32 chat template (#9814) 
Signed-off-by: Chang Liu (Enterprise Products) <9713593+chang-l@users.noreply.github.com>
 2025-12-19 17:05:38 +08:00
William Zhang
478b6b20a1
[#9230][refactor] Replace nemotron patches with custom model implementation (#9751) 
[#9230][refactor] Replace nemotron patches with custom model implementation

* Why?

Patching for nemotron H models was growing out of hand, and made certain
optimizations more complex than they needed to be.

* What?

This commit finally gets rid of them, and replaces them with the custom
model implementation in `modeling_nemotron_h.py`.

Closes #9230
Closes NvBug 5747867

Signed-off-by: William Zhang <133824995+2ez4bz@users.noreply.github.com>
 2025-12-18 19:36:27 -08:00
Ivy Zhang
9aa40871c2
[TRTLLM-9840][test] switch ucx backend to default backend (#10101) 
Signed-off-by: Ivy Zhang <25222398+crazydemo@users.noreply.github.com>
 2025-12-18 18:54:15 -08:00
Anish Shanbhag
91a9ae42d2
[TRTC-71][feat] Add regression testing for config database (#9832) 
Signed-off-by: Anish Shanbhag <ashanbhag@nvidia.com>
 2025-12-18 16:15:38 -08:00
Chang Liu
a97e411b44
[https://nvbugs/5747911][fix] Use offline data path for the unit test of mmencoder server (#10135) 
Signed-off-by: Chang Liu (Enterprise Products) <9713593+chang-l@users.noreply.github.com>
 2025-12-18 15:19:23 -08:00
Lizhi Zhou
f02782a6f2
[https://nvbugs/5726066][fix] fix auto-scaling related failures (#9845) 
Signed-off-by: Lizhi Zhou <1432185+reasonsolo@users.noreply.github.com>
Co-authored-by: Emma Qiao <qqiao@nvidia.com>
 2025-12-18 16:37:48 -05:00
CarstyYou
0b279f4ad4
[https://nvbugs/5456493][feat] Add fp8 bmm on sm120 (#9687) 
Signed-off-by: CarstyYou <186021327+CarstyYou@users.noreply.github.com>
 2025-12-18 22:57:20 +08:00
ZhichenJiang
4e55b83101
[None][perf] Add more optimization options for MOE CuteDSL finalized kernel (#10042) 
Signed-off-by: zhichen jiang <zhichenj@NVIDIA.com>
 2025-12-18 22:49:28 +08:00