All pull requests

Pull requests

• [None][infra] Waive 2 failed cases for main in pre-merge 66681

  #21141·trtllm-agent opened Oct 28, 2026Collaborator·

  ·
  2

  

• [None][infra] Waive 1 failed cases for main in pre-merge 66696

  #21140·trtllm-agent opened Oct 28, 2026Collaborator·

  ·
  3

  

• [https://nvbugs/6907190][chore] Unwaive test_llm_update_weights_qwen35_35b_bf16

  #21135·hchings opened Oct 27, 2026Collaborator·1 task·

  ·
  3

  

• [TRTLLM-17383][infra] Download Boost from AWS Artifactory

  #21134·sharabeshj-nvda opened Oct 27, 2026Collaborator·1 task done·

  ·
  3

  

• [TRTLLM-17383][infra] Download Boost from AWS Artifactory

  ci: full pre-merge approved
  #21133·sharabeshj-nvda opened Oct 27, 2026Collaborator·1 task done·

  ·
  4

  

• [None][perf] Avoid output concatenation in eager GDN

  ci: full pre-merge approved
  #21132·2ez4bz opened Oct 27, 2026Collaborator·1 task done·

  ·
  7

  

• [TRTLLM-17359][fix] Preserve per-rank attention-DP KV metrics

  #21131·chienchunhung opened Oct 27, 2026Collaborator·1 task done·

  ·
  1

  6

  

• [None][fix] Compile the DSpark non-causal block merge and size its LSE workspace

  #21130·zheyuf opened Oct 27, 2026Collaborator·1 task·

  ·
  

• [None][feat] KVCM2 core: sparse attention arena, host-backed pools and Batch

  #21129·lowsfer opened Oct 27, 2026Member·1 task done·

  ·
  19

  

• [None][fix] revert git pin in 8f04d3a

  #21128·tburt-nv opened Oct 27, 2026Collaborator·1 task done·

  ·
  4

  

• [None][infra] Add retry for Build stage

  #21127·yuanjingx87 opened Oct 27, 2026Collaborator·1 task·

  ·
  4

  

• [None][perf] SID-GR 4/N: Skip beam search temperature pass when temperature is 1

  ci: full pre-merge approved
  #21126·mikeiovine opened Oct 27, 2026Collaborator·1 task done·

  ·
  6

  

• [None][feat] Add image routing metadata to OpenEngine KV events

  #21125·krishung5 opened Oct 27, 2026Collaborator·1 task done·

  ·
  14

  

• [None][test] Add a warmup pass before perf-sanity measured clients

  #21123·chenfeiz0326 opened Oct 27, 2026Collaborator·1 task done·

  ·
  5

  

• [TRTLLM-17125][feat] Preempt requests on KV pressure under aggregated Helix

  #21122·asfiyab-nvidia opened Oct 27, 2026Contributor·1 task done·

  ·
  12

  

• [https://nvbugs/6993575][fix] Prime MegaMoE tile variants during autotuning

  ci: full pre-merge approved
  #21121·trtllm-agent opened Oct 27, 2026Collaborator·

  ·
  2

  

• [None][fix] KVCM v2: plan drops past private SWA layers; hold the manager in prefetch_reuse

  #21120·chenfeiz0326 opened Oct 27, 2026Collaborator·1 task done·

  ·
  1

  

• [None][perf] KVCacheV2Scheduler: optionally separate context and generation passes

  #21119·chenfeiz0326 opened Oct 27, 2026Collaborator·1 task done·

  ·
  3

  

• [None][perf] Disagg gen_only_no_context: do not wait on receive fences at admission

  #21118·chenfeiz0326 opened Oct 27, 2026Collaborator·1 task done·

  ·
  3

  

• [None][perf] Bit-exact warp-per-row kernel for the non-fused MoE finalize

  Community want to contribute
  #21116·Heisenberg-Yin opened Oct 27, 2026·1 task done·

  ·
  1

  

• [None][feat] Add a global memory monitor

  #21113·zhaoyangwang-nvidia opened Oct 27, 2026Collaborator·1 task done·

  ·
  

• [None][fix] Kimi-K3 parsers: one reading for stream and final response, and read calls with missing close tags

  #21112·JunyiXu-nv opened Oct 27, 2026Collaborator·1 task done·

  ·
  4

  

• [None][fix] serve: take Kimi-K3 chat prompt ids from its renderer; count_tokens matches usage

  #21111·JunyiXu-nv opened Oct 27, 2026Collaborator·1 task done·

  ·
  12

  

• [None][fix] GLM reasoning parser: drop the owed </think> after a tool call; read the thinking mode off the prompt

  #21109·JunyiXu-nv opened Oct 27, 2026Collaborator·1 task done·

  ·
  12

  

• [None][fix] GLM tool parsers: a fresh call in an open call's arguments abandons it

  #21108·JunyiXu-nv opened Oct 27, 2026Collaborator·1 task done·

  ·
  11