Commits

Branch selector

 main

User selector

All users

Datepicker

All time

Commit History

Commits on Dec 19, 2024

• Set the url to pytorch/ao and clean up some duplication (pytorch#1446)

  janeyx99

  authored
  aea2356

  Copy full SHA for aea2356

• [2/2] Switch ao to use py_limited_api (pytorch#1277)

  Show description for 1beb6d0
  janeyx99

  authored
  1beb6d0

  Copy full SHA for 1beb6d0

• fix local float8 tests on H100 (pytorch#1438)

  Show description for 38c79d4
  vkuzo

  authored
  38c79d4

  Copy full SHA for 38c79d4

Commits on Dec 18, 2024

• metal lowbit kernels: optimized 2-bit, 3-bit and 4-bit shaders (pytorch#1422)

  manuelcandales

  authored
  2e032c6

  Copy full SHA for 2e032c6

• [resubmit] Gemlite fix (pytorch#1435)

  Show description for f52d3ab
  jerryzh168

  authored
  f52d3ab

  Copy full SHA for f52d3ab

• [float8nocompile] Simplified Float8Linear implementation which only supports dynamic tensorwise scaling (pytorch#1429)

  Show description for ec64182
  

  danielvegamyhre

  and

  Daniel Vega-Myhre

  authored
  ec64182

  Copy full SHA for ec64182

• Fixes observer attachment to model based on config for wanda sparsifier (pytorch#1265)

  Show description for 33d57af
  agrawal-aka

  authored
  33d57af

  Copy full SHA for 33d57af

• [ez] Use noquant to make dashboard logic easier (pytorch#1430)

  Show description for a03ca99
  jerryzh168

  authored
  a03ca99

  Copy full SHA for a03ca99

Commits on Dec 17, 2024

• Add local optoin for output json file (pytorch#1426)

  Show description for a5a53a2
  jerryzh168

  authored
  a5a53a2

  Copy full SHA for a5a53a2

• [float8nocompile] Initial commit of float8nocompile prototype (pytorch#1428)

  Show description for 02d413c
  

  danielvegamyhre

  and

  Daniel Vega-Myhre

  authored
  02d413c

  Copy full SHA for 02d413c

• [WIP] Codebook quantization flow (pytorch#1299)

  Show description for bc000aa
  DerekLiu35

  authored
  bc000aa

  Copy full SHA for bc000aa

• Fix error message in float8 FSDP utils (pytorch#1423)

  Show description for ae28114
  

  danielvegamyhre

  and

  Daniel Vega-Myhre

  authored
  ae28114

  Copy full SHA for ae28114

• Add dashboard metric support for sam and sam2 benchmarks (pytorch#1407)

  Show description for cbd7c29
  jerryzh168

  authored
  cbd7c29

  Copy full SHA for cbd7c29

• Expose zero_point_domain as arguments (pytorch#1401)

  Show description for ace7219
  

  airMeng

  and

  ruff

  authored
  ace7219

  Copy full SHA for ace7219

Commits on Dec 16, 2024

• Move Int4CPULayout to int4_cpu_layout.py (pytorch#1419)

  Show description for 9472a17
  

  yanbing-j

  and

  RUFF-bot

  authored
  9472a17

  Copy full SHA for 9472a17

• gemlite integration in torchao (pytorch#1034)

  Show description for 603d908
  HDCharles

  authored
  603d908

  Copy full SHA for 603d908

• Add new QAT API through quantize_ (pytorch#1415)

  Show description for 200589b
  

  andrewor14

  authored and

  pytorchmergebot

  committed
  200589b

  Copy full SHA for 200589b

Commits on Dec 13, 2024

• Add back mistakenly deleted QAT BC import test (pytorch#1417)

  Show description for 46b8796
  andrewor14

  authored
  46b8796

  Copy full SHA for 46b8796

• Clarify how users should fix ruff errors (pytorch#1416)

  Show description for a555734
  andrewor14

  authored
  a555734

  Copy full SHA for a555734

• metal lowbit kernels: glob metal files in CMakeLists (pytorch#1410)

  manuelcandales

  authored
  7d7c14e

  Copy full SHA for 7d7c14e

• metal lowbit kernels: executorch ops

  Show description for ebc4303
  manuelcandales

  authored
  ebc4303

  Copy full SHA for ebc4303

Commits on Dec 12, 2024

• Update index.rst (pytorch#1409)

  jerryzh168

  authored
  31234db

  Copy full SHA for 31234db

• Update api_ref_quantization.rst (pytorch#1408)

  jerryzh168

  authored
  eed437f

  Copy full SHA for eed437f

• Enable ciflow/benchmark (pytorch#1404)

  Show description for 7624ae8
  huydhn

  authored
  7624ae8

  Copy full SHA for 7624ae8

• SAM2 video batching (pytorch#1395)

  cpuhrsch

  authored
  19b3bb5

  Copy full SHA for 19b3bb5

Commits on Dec 11, 2024

• Add marlin and semi sparse + quant option to autoquant (pytorch#1399)

  Show description for 039cef4
  jerryzh168

  authored
  039cef4

  Copy full SHA for 039cef4

• Fix a bug in LinearActivationQuantizedTensor (pytorch#1400)

  Show description for 63b30ca
  jerryzh168

  authored
  63b30ca

  Copy full SHA for 63b30ca

• Add exhaustive config option to intmm kernel (pytorch#1392)

  Show description for cac5261
  jerryzh168

  authored
  cac5261

  Copy full SHA for cac5261

Commits on Dec 10, 2024

• Add llama benchmark run to GHA (pytorch#1398)

  Show description for f258d82
  

  huydhn

  and

  jerryzh168

  authored
  f258d82

  Copy full SHA for f258d82

Commits on Dec 9, 2024

• Update SAM AMG README with more descriptive install instructions (pytorch#1337)

  philipbutler

  authored
  a6f8676

  Copy full SHA for a6f8676

Commits on Dec 7, 2024

• Move MarlinQQQTensor out of AQT (pytorch#1385)

  jainapurva

  authored
  23db9bf

  Copy full SHA for 23db9bf

Commits on Dec 6, 2024

• SAM2: video_profile.py (pytorch#1386)

  cpuhrsch

  authored
  8a805d0

  Copy full SHA for 8a805d0

Commits on Dec 5, 2024

• Add script to pre-process raw release notes (pytorch#1380)

  Show description for abff563
  andrewor14

  authored
  abff563

  Copy full SHA for abff563

Commits on Dec 4, 2024

• [float8] Allow specifying arbitrary dtype for each tensor (pytorch#1378)

  lw

  authored
  4f8021f

  Copy full SHA for 4f8021f

• [float8] Re-enable slow-accum in the bwd of axis-wise scaling schemes (pytorch#1377)

  lw

  authored
  6a177c9

  Copy full SHA for 6a177c9

Pagination

Previous

Next