Commits

Branch selector

main

User selector

All users

All time

Commit history

Commits on Jun 12, 2024

• Update version to 0.3.0 (#348)

  Show description for eae864c
  

  Mark Saroufim

  authored
  eae864cView commit details

  Copy full SHA for eae864c

  Browse repository at this point

Commits on Jun 11, 2024

• task_list to tasks (#343)

  HDCharles

  authored
  950a893View commit details

  Copy full SHA for 950a893

  Browse repository at this point

• Adding Llama to TorchAO (#276)

  Show description for 61fef69
  HDCharles

  authored
  61fef69View commit details

  Copy full SHA for 61fef69

  Browse repository at this point

• Improving hf_eval.py (#342)

  Show description for 71cf0a5
  HDCharles

  authored
  71cf0a5View commit details

  Copy full SHA for 71cf0a5

  Browse repository at this point

• fix compile (#341)

  gau-nernst

  authored
  7cddc6eView commit details

  Copy full SHA for 7cddc6e

  Browse repository at this point

Commits on Jun 10, 2024

• Fix dimension issues for int4 weight only quant path (#330)

  Show description for 79f2c7f
  jerryzh168

  authored
  79f2c7fView commit details

  Copy full SHA for 79f2c7f

  Browse repository at this point

• Bitpackingv2 (#307)

  Show description for c38fc4a
  vayuda

  authored
  c38fc4aView commit details

  Copy full SHA for c38fc4a

  Browse repository at this point

• FP6-LLM clean up (again) (#339)

  gau-nernst

  authored
  6a380a3View commit details

  Copy full SHA for 6a380a3

  Browse repository at this point

• Update regression test to python 3.8 (#340)

  

  Mark Saroufim

  authored
  0d44041View commit details

  Copy full SHA for 0d44041

  Browse repository at this point

Commits on Jun 9, 2024

• Revert "Clean up FP6-LLM" (#338)

  Show description for 59e57db
  

  Mark Saroufim

  authored
  59e57dbView commit details

  Copy full SHA for 59e57db

  Browse repository at this point

• Move quant ops to utils.py (#331)

  Show description for 8c8bc81
  jerryzh168

  authored
  8c8bc81View commit details

  Copy full SHA for 8c8bc81

  Browse repository at this point

• Clean up FP6-LLM (#304)

  Show description for cd8f647
  gau-nernst

  authored
  cd8f647View commit details

  Copy full SHA for cd8f647

  Browse repository at this point

Commits on Jun 7, 2024

• Move some util functions from quantization.utils to torchao.utils (#337)

  Show description for 000a0fd
  jerryzh168

  authored
  000a0fdView commit details

  Copy full SHA for 000a0fd

  Browse repository at this point

• Add architecture to index postfix for nightly builds (#336)

  

  Mark Saroufim

  authored
  335171fView commit details

  Copy full SHA for 335171f

  Browse repository at this point

• Update nightly build instructions (#334)

  

  Mark Saroufim

  authored
  c8d9fa5View commit details

  Copy full SHA for c8d9fa5

  Browse repository at this point

• Unpin CUDA nightly (#333)

  Show description for ff42c85
  cpuhrsch

  authored
  ff42c85View commit details

  Copy full SHA for ff42c85

  Browse repository at this point

Commits on Jun 6, 2024

• training acceleration via runtime semi-structured sparsity (#184)

  Show description for d97ae74
  

  Jesse Cai

  authored
  d97ae74View commit details

  Copy full SHA for d97ae74

  Browse repository at this point

• Add `AUTOQUANT_CACHE` docs for reusing the same quantization plan (#329)

  RobinKa

  authored
  f3f2ea8View commit details

  Copy full SHA for f3f2ea8

  Browse repository at this point

Commits on Jun 5, 2024

• Refactor rest of tinygemm quant primitive ops (#321)

  Show description for 03e2c9b
  jerryzh168

  authored
  03e2c9bView commit details

  Copy full SHA for 03e2c9b

  Browse repository at this point

• [NF4Tensor] Switch to save for backward since are now a tensor input (#323)

  drisspg

  authored
  08fb8bfView commit details

  Copy full SHA for 08fb8bf

  Browse repository at this point

• Wrap torch.ops.quantized_decomposed to improve import errors (#310)

  cpuhrsch

  authored
  1518f7dView commit details

  Copy full SHA for 1518f7d

  Browse repository at this point

Commits on Jun 4, 2024

• Adding a quick way for users to test model eval for hf models (#328)

  HDCharles

  authored
  98d7c3dView commit details

  Copy full SHA for 98d7c3d

  Browse repository at this point

• Pin CUDA nightly to mitigate regression (#322)

  Show description for e2196fd
  cpuhrsch

  authored
  e2196fdView commit details

  Copy full SHA for e2196fd

  Browse repository at this point

• Refactor int4 and int8 weight only quantization to use `quantize` (#301)

  Show description for 338d87c
  

  jerryzh168

  and

  Mark Saroufim

  authored
  338d87cView commit details

  Copy full SHA for 338d87c

  Browse repository at this point

• Add fused QKV `HQQ` `triton_mm` test (#306)

  jeromeku

  authored
  729fa4dView commit details

  Copy full SHA for 729fa4d

  Browse repository at this point

• Fix FP6-LLM benchmark (#312)

  gau-nernst

  authored
  8dbf031View commit details

  Copy full SHA for 8dbf031

  Browse repository at this point

• Update old pretrained TorchVision API in ao tutorials (#313) (#314)

  Show description for 12f44ab
  kit1980

  authored
  12f44abView commit details

  Copy full SHA for 12f44ab

  Browse repository at this point

Commits on Jun 3, 2024

• Check dequantize_affine is idempotent (#309)

  cpuhrsch

  authored
  d75f450View commit details

  Copy full SHA for d75f450

  Browse repository at this point

• Retry of D58015187 Move AsyncCompile to a different file

  Show description for 88daa1a
  jamesjwu

  authored
  88daa1aView commit details

  Copy full SHA for 88daa1a

  Browse repository at this point

• sparse benchmarking numbers (#303)

  Show description for e5d27a3
  

  Jesse Cai

  authored
  e5d27a3View commit details

  Copy full SHA for e5d27a3

  Browse repository at this point

Commits on Jun 1, 2024

• Re-enable AOTI tests (#212)

  jerryzh168

  authored
  8a4e693View commit details

  Copy full SHA for 8a4e693

  Browse repository at this point

Commits on May 31, 2024

• [NF4][FSDP] return contiguous `quantization_factor` (#298)

  Show description for 9dbc554
  weifengpy

  authored
  9dbc554View commit details

  Copy full SHA for 9dbc554

  Browse repository at this point

• Refactor int8 dynamic quantization with call to `quantize` (#294)

  Show description for 68ce5b8
  jerryzh168

  authored
  68ce5b8View commit details

  Copy full SHA for 68ce5b8

  Browse repository at this point

Commits on May 30, 2024

• Graceful handling of cpp extensions (#296)

  Show description for e7837d7
  

  Mark Saroufim

  authored
  e7837d7View commit details

  Copy full SHA for e7837d7

  Browse repository at this point

• Fixed f-string printing of `NF4Tensor`s (#297)

  awgu

  authored
  4c1d568View commit details

  Copy full SHA for 4c1d568

  Browse repository at this point