Code

Add fp4 support(WIP)#3496

Code

Closed

lanluo-nvidia wants to merge 15 commits intomainpytorch/TensorRT:mainfrom

lluo/fp4pytorch/TensorRT:lluo/fp4Copy head branch name to clipboard

Commits

Commits on Apr 29, 2025

• Add fp4 support

  lanluo-nvidia

  committed
  1d172ce

  Copy full SHA for 1d172ce

Commits on Apr 30, 2025

• test

  lanluo-nvidia

  committed
  d2b1422

  Copy full SHA for d2b1422

Commits on May 1, 2025

• Merge branch 'main' into lluo/fp4

  lanluo-nvidia

  committed
  38617b4

  Copy full SHA for 38617b4

• upgrade modelopt

  lanluo-nvidia

  committed
  d439d96

  Copy full SHA for d439d96

Commits on May 2, 2025

• add constant fold

  lanluo-nvidia

  committed
  5a2213e

  Copy full SHA for 5a2213e

• fix the input tensor type issue

  lanluo-nvidia

  committed
  fcf0c12

  Copy full SHA for fcf0c12

Commits on May 5, 2025

• test

  lanluo-nvidia

  committed
  057f35a

  Copy full SHA for 057f35a

Commits on May 6, 2025

• test

  lanluo-nvidia

  committed
  7b09862

  Copy full SHA for 7b09862

• test

  lanluo-nvidia

  committed
  d9f2ad9

  Copy full SHA for d9f2ad9

• test

  lanluo-nvidia

  committed
  6892a47

  Copy full SHA for 6892a47

Commits on May 7, 2025

• Merge branch 'main' into lluo/fp4

  lanluo-nvidia

  committed
  5198f9a

  Copy full SHA for 5198f9a

• restructure the dynamic double quantize and static double quantize code

  lanluo-nvidia

  committed
  559ada5

  Copy full SHA for 559ada5

Commits on May 14, 2025

• add test code

  lanluo-nvidia

  committed
  bba1d79

  Copy full SHA for bba1d79

• test

  lanluo-nvidia

  committed
  f16e58a

  Copy full SHA for f16e58a

• test

  lanluo-nvidia

  committed
  06c8126

  Copy full SHA for 06c8126