-
Notifications
You must be signed in to change notification settings - Fork 2.7k
Pull requests: NVIDIA/TensorRT-LLM
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
[TRTLLM-14968][TRTLLM-14969][refactor] Split TRTLLMGenFusedMoE into eleven leaves with canonical identities
#19153
opened Sep 14, 2026 by
xxi-nv
Collaborator
Loading…
1 task done
[TRTLLMINF-419][infra] Clean up the devel image build
#19152
opened Sep 14, 2026 by
EmmaQiaoCh
Collaborator
•
Draft
1 task
[None][fix] Pass alpha through SM107 scaled_mm compilation
#19150
opened Sep 14, 2026 by
peaceh-nv
Collaborator
Loading…
[None][feat] add powerlaw expert_pattern to bench_moe routing control
#19148
opened Sep 14, 2026 by
leslie-fang25
Collaborator
Loading…
1 task done
[#18083][fix] Resolve Qwen3 reasoning mode from the rendered prompt
#19147
opened Sep 14, 2026 by
somtri
Loading…
1 task done
[None][chore] Name the Nemotron multimodal module for the family it serves instead of one of its three models
#19146
opened Sep 14, 2026 by
eopXD
Collaborator
Loading…
1 task done
[None][feat] Add opt-in subprocess profiling for split-K tactics
#19145
opened Sep 14, 2026 by
peaceh-nv
Collaborator
Loading…
1 task done
[https://nvbugs/6716104][fix] Keep mixed BF16 attention on FlashInfer
#19144
opened Sep 14, 2026 by
yihwang-nv
Collaborator
Loading…
1 task done
[None][fix] Reject unsupported SM100 sync-object factories
#19142
opened Sep 14, 2026 by
peaceh-nv
Collaborator
Loading…
1 task done
[TRTLLM-16220][feat] Enable KVCacheManagerV2 by default for Nemotron multimodal models
#19140
opened Sep 14, 2026 by
eopXD
Collaborator
Loading…
1 task done
[None][fix] Complete DeepSeek-V4 Rubin BF16 dispatch and optimize MLA KV expansion
#19139
opened Sep 14, 2026 by
peaceh-nv
Collaborator
Loading…
[None][feat] enable the GVR Top-K and MTP draft-loop index reuse for the QSA indexer
api-compatible
Accepted LLM API contract change that is backwards-compatible
#19138
opened Sep 14, 2026 by
Wanli-Jiang
Collaborator
Loading…
[None][feat] Add GLM-5.3-Flash (glm5_next) support on shared modules
#19136
opened Sep 14, 2026 by
ruocheng-nv
Loading…
1 task done
[None][fix] Give each VisualGen server its own media directory
VisualGen
#19135
opened Sep 14, 2026 by
luyiyun1021
Collaborator
Loading…
1 task done
[None][feat] Make ADP new conversation routing token-aware
#19130
opened Sep 14, 2026 by
dongfengy
Collaborator
Loading…
1 task
[https://nvbugs/6770503][test] Update perf sanity helper unit tests to match #18990 renames
ci: full pre-merge approved
#19129
opened Sep 14, 2026 by
HuiGao-NV
Collaborator
Loading…
1 task done
[TRTLLM-15715][refactor] Extract progress polling and error consensus into DisaggTransferCoordinator
ci: full pre-merge approved
#19128
opened Sep 14, 2026 by
nv-xtf
Collaborator
Loading…
1 task done
[TRTLLMINF-417][infra] Reject explicit 60-minute CI test-list timeouts
#19126
opened Sep 14, 2026 by
chzblych
Collaborator
Loading…
1 task done
[TRTLLM-16315][infra] add Python change reference analysis for CBTS
#19124
opened Sep 14, 2026 by
crazydemo
Collaborator
Loading…
1 task done
[None][fix] layer-wise calibrator: resolve the decoder stack instead of assuming model.model.layers
#19122
opened Sep 14, 2026 by
ruodil
Collaborator
Loading…
3 tasks done
[#18661][fix] Keep KVCacheManagerV2 resize capacity >= history under MTP rewind
#19120
opened Sep 13, 2026 by
dundysm
Loading…
3 of 4 tasks
[None][fix] Skip sampling after absorbed PyTorch forward step
#19118
opened Sep 13, 2026 by
lorenzozanee
Loading…
[#18847][fix] Restore worker CPU affinity at shutdown
VisualGen
#19117
opened Sep 13, 2026 by
DOCCA0
Loading…
1 task done
Previous Next
ProTip!
Adding no:label will show everything without a label.