All pull requests

Pull requests

• Key the NemotronH NVFP4 PTQ recipe to the Nemotron 3.5 Super VL checkpoint

  #2819·juhi10071998 opened Oct 15, 2026Contributor·

  ·
  1

• docs: announce Nemotron 3.5 Super VL NVFP4 checkpoint

  #2818·Edwardf0t1 opened Oct 15, 2026Contributor·

  ·
  1

• Fix MiniMax M2.7 expert-only recipe for modern Transformers

  #2817·chadvoegele opened Oct 15, 2026Contributor·

  ·
  3

• Fix Windows ONNX external data corruption by upgrading ONNX

  #2816·anmaurya001 opened Oct 15, 2026·

  ·
  5

• [6933164] fix: preserve Diffusers pipeline settings in component exports

  cherry-pick-0.48.0
  #2813·yingguo-trt opened Oct 15, 2026Contributor·

  ·
  7

• Upgrade ONNX Runtime 1.30 and CUDA 13 for Linux x86-64

  cherry-pick-0.48.0
  #2809·ajrasane opened Oct 14, 2026Contributor·

  ·
  11

• [6918844] Fix LSQ scale floor

  cherry-pick-0.48.0
  #2808·rohansjoshi opened Oct 14, 2026Contributor·

  ·
  5

• Fix fused-experts export for W4A8 AWQ (SequentialQuantizer) weight quantizers

  cherry-pick-0.48.0
  #2807·Fridah-nv opened Oct 14, 2026Contributor·

  ·
  16

• [4/8] Add dynamic NVFP4 recurrent-state QAT

  #2805·kaix-nv opened Oct 14, 2026Contributor·3/7·

  ·
  4

• docs(eval): pin the LiveCodeBench judge to fix binary-stdio grading

  #2800·cjluo-nv opened Oct 14, 2026Collaborator·

  ·
  5

• [6915422] Fix TensorRT 11 inference shape validation

  cherry-pick-0.48.0
  #2798·ajrasane opened Oct 14, 2026Contributor·

  ·
  9

• Let the multimodal generator enable thinking and fit its output budget

  #2797·yeyu-nvidia opened Oct 14, 2026Contributor·

  ·
  7

• [6915426] Fix ONNX AutoTune benchmark initialization on TensorRT 11

  cherry-pick-0.48.0
  #2795·ajrasane opened Oct 14, 2026Contributor·

  ·
  7

• [5/8] ReplaySSM state QAT and two-precision GDN

  #2788·kaix-nv opened Oct 14, 2026Contributor·4/7·

  ·
  4

• Make freeze_mtp an explicit override of the automatic MTP freeze

  #2784·yueshen2016 opened Oct 13, 2026Contributor·

  ·
  6

• [OMNIML-6060] [12/21] Load an expert-parallel plan through load_pretrained

  #2780·shengliangxu opened Oct 13, 2026Collaborator·13/22·

  ·
  7

• [OMNIML-6054] [6/21] Route load_pretrained by the distribution plan

  #2779·shengliangxu opened Oct 13, 2026Collaborator·7/22·

  ·
  11

• [2/2] Unified export: quantized-weight packing and state transforms

  cherry-pick-0.48.0export/deploy
  #2778·mxinO opened Oct 13, 2026Contributor·

  ·
  11

• [1/2] Unified export: quantized-weight state capture

  cherry-pick-0.48.0export/deploy
  #2777·mxinO opened Oct 13, 2026Contributor·

  ·
  8

• Sync GPTQ Hessians across data-parallel ranks

  cherry-pick-0.48.0
  #2773·kevalmorabia97 opened Oct 13, 2026Collaborator·

  ·
  12

• [https://nvbugspro.nvidia.com/bug/6918844] Fix LSQ FP8 block-scale lower bound

  cherry-pick-0.48.0
  #2772·yingguo-trt opened Oct 13, 2026Contributor·

  ·
  3

  

• [NVBug -6914213] fix(export): drop reverse renames that invent names the source checkpoint lacks

  cherry-pick-0.48.0
  #2766·juhi10071998 opened Oct 12, 2026Contributor·

  ·
  48

• fix(speculative): DSpark and Domino AR validation and label alignment

  #2762·h-guo18 opened Oct 12, 2026Contributor·

  ·
  2

• fix(speculative): keep padding out of DSpark and Domino loss masks

  #2761·h-guo18 opened Oct 12, 2026Contributor·

  ·
  3

• Do not leave 'trt' in the autocast providers default after one conversion

  #2760·v0ropaev opened Oct 12, 2026Contributor·

  ·
  2