Commits

Branch selector

 main

User selector

andrewor14

Datepicker

All time

Commit History

Commits on Nov 26, 2025

• Fix fbcode loading cpp extensions (#3372)

  Show description for 38d55dc
  andrewor14

  authored
  38d55dc

  Copy full SHA for 38d55dc

• Fix style after https://github.com/pytorch/ao/pull/3261 (#3397)

  andrewor14

  authored
  316ef03

  Copy full SHA for 316ef03

• Unbreak BC, add back PerRow, PerTensor imports (#3396)

  Show description for eb2a8fc
  andrewor14

  authored
  eb2a8fc

  Copy full SHA for eb2a8fc

Commits on Nov 25, 2025

• Add conv2d support for IntxUnpackedToInt8Tensor (#3371)

  Show description for 4903c55
  andrewor14

  authored
  4903c55

  Copy full SHA for 4903c55

Commits on Nov 23, 2025

• Add PerBlock to safe globals (#3370)

  Show description for b55713a
  andrewor14

  authored
  b55713a

  Copy full SHA for b55713a

Commits on Nov 14, 2025

• [pt2e] Fix QAT annotations for special qspecs (#3337)

  Show description for 8496b55
  andrewor14

  authored
  8496b55

  Copy full SHA for 8496b55

Commits on Nov 10, 2025

• Remove old TORCH_VERSION variables (#3146)

  Show description for e8c4d09
  andrewor14

  authored
  e8c4d09

  Copy full SHA for e8c4d09

Commits on Nov 7, 2025

• Update Float8Tensor for GRPO training in unsloth (#3158)

  Show description for 6259e98
  andrewor14

  authored
  6259e98

  Copy full SHA for 6259e98

Commits on Oct 31, 2025

• Update torchao + unsloth integration on README (#3267)

  Show description for e418066
  andrewor14

  authored
  e418066

  Copy full SHA for e418066

Commits on Oct 23, 2025

• Add Unsloth + QAT blog to latest news (#3227)

  andrewor14

  authored
  f3fc5e7

  Copy full SHA for f3fc5e7

Commits on Oct 20, 2025

• Update QAT README before PTC (#3214)

  Show description for ad533f5
  andrewor14

  authored
  ad533f5

  Copy full SHA for ad533f5

Commits on Oct 17, 2025

• Fix typo in AQT test (#3205)

  andrewor14

  authored
  e30012f

  Copy full SHA for e30012f

Commits on Oct 16, 2025

• Revert "Remove config functions like `int4_weight_only` (#3145)" (#3192)

  Show description for 8e63472
  andrewor14

  authored
  8e63472

  Copy full SHA for 8e63472

• Fix TORCHAO_SKIP_LOADING_SO_FILES behavior (#3189)

  Show description for 01232b9
  andrewor14

  authored
  01232b9

  Copy full SHA for 01232b9

• Update compatibility matrix (#3178)

  Show description for 16c7d09
  andrewor14

  authored
  16c7d09

  Copy full SHA for 16c7d09

Commits on Oct 13, 2025

• Remove config functions like `int4_weight_only` (#3145)

  Show description for 53a66f8
  andrewor14

  authored
  53a66f8

  Copy full SHA for 53a66f8

Commits on Oct 10, 2025

• Fix pt2e test_qat_preserve_source_fn_stack (#3149)

  Show description for fa17427
  andrewor14

  authored
  fa17427

  Copy full SHA for fa17427

Commits on Sep 30, 2025

• Improve QAT nvfp4 numerics (#3050)

  Show description for d407246
  andrewor14

  authored
  d407246

  Copy full SHA for d407246

Commits on Sep 25, 2025

• Support NVFP4 dynamic per tensor scale (#3049)

  Show description for a79e66c
  andrewor14

  authored
  a79e66c

  Copy full SHA for a79e66c

Commits on Sep 19, 2025

• Pass QAT learned qparams in convert (#3022)

  Show description for ae12e42
  andrewor14

  authored
  ae12e42

  Copy full SHA for ae12e42

Commits on Sep 17, 2025

• Remove internal usage of all config functions like `int4_weight_only` (#2995)

  Show description for 9e5059e
  andrewor14

  authored
  9e5059e

  Copy full SHA for 9e5059e

• Deprecate config functions like `int4_weight_only` (#2994)

  Show description for 62f62d0
  andrewor14

  authored
  62f62d0

  Copy full SHA for 62f62d0

Commits on Sep 16, 2025

• Rename Int4WeightPreshuffledFakeQuantizeConfig (#3005)

  Show description for 58c3064
  andrewor14

  authored
  58c3064

  Copy full SHA for 58c3064

Commits on Sep 15, 2025

• Improve QAT int4 weight-only numerics (#2986)

  Show description for ea8c00f
  andrewor14

  authored
  ea8c00f

  Copy full SHA for ea8c00f

Commits on Sep 11, 2025

• [pt2e] Make prepare and convert faster by caching (#2983)

  Show description for be71434
  andrewor14

  authored
  be71434

  Copy full SHA for be71434

Commits on Sep 9, 2025

• Skip QAT tests using `quantize_fp8_row` in fbcode (#2963)

  Show description for 10ba659
  andrewor14

  authored
  10ba659

  Copy full SHA for 10ba659

Commits on Sep 8, 2025

• Fix ROCM QAT test failure (#2957)

  Show description for 2ccab32
  andrewor14

  authored
  2ccab32

  Copy full SHA for 2ccab32

• Skip expanding scales for rowwise fp8 quantize (#2950)

  Show description for a54417d
  andrewor14

  authored
  a54417d

  Copy full SHA for a54417d

• Skip QAT int4 v2 test for fbcode (#2923)

  Show description for e368b61
  andrewor14

  authored
  e368b61

  Copy full SHA for e368b61

• Improve QAT fp8-int4 numerics (#2937)

  Show description for a2206e9
  andrewor14

  authored
  a2206e9

  Copy full SHA for a2206e9

Commits on Sep 3, 2025

• [pt2e] Avoid getting model device once per node (#2695)

  Show description for 870284f
  andrewor14

  authored
  870284f

  Copy full SHA for 870284f

Commits on Aug 29, 2025

• Remove unused cpp variable, breaking style checks (#2909)

  Show description for 1bb1a40
  andrewor14

  authored
  1bb1a40

  Copy full SHA for 1bb1a40

Commits on Aug 28, 2025

• Support QAT int4 v1 path for BC (#2888)

  Show description for 6e9bf26
  andrewor14

  authored
  6e9bf26

  Copy full SHA for 6e9bf26

Commits on Aug 25, 2025

• Add NVFP4 QAT (#2666)

  Show description for f3e549c
  andrewor14

  authored
  f3e549c

  Copy full SHA for f3e549c

Commits on Aug 22, 2025

• Fix test_nvfp4_tensor.py merge conflict (#2857)

  Show description for 8079abc
  andrewor14

  authored
  8079abc

  Copy full SHA for 8079abc

Pagination

Previous

Next