Commits

Branch selector

 gh/liangel/1/head

User selector

All users

Datepicker

All time

Commit History

Commits on Sep 25, 2025

• Update on "add int4tensor support for safetensors"

  Show description for f7b8068
  liangel-02

  committed
  f7b8068

  Copy full SHA for f7b8068

• Update base for Update on "add int4tensor support for safetensors"

  Show description for 512f9ed
  liangel-02

  committed
  512f9ed

  Copy full SHA for 512f9ed

• [Reland] Unify get_block_size (#3059)

  Show description for 0064084
  Xia-Weiwen

  authored
  0064084

  Copy full SHA for 0064084

• [CPU] Add ops for float8 linear (#3052)

  Show description for 5e90c47
  Xia-Weiwen

  authored
  5e90c47

  Copy full SHA for 5e90c47

Commits on Sep 24, 2025

• Update on "add int4tensor support for safetensors"

  Show description for 86d7fcf
  liangel-02

  committed
  86d7fcf

  Copy full SHA for 86d7fcf

• Update on "add int4tensor support for safetensors"

  Show description for 6e31683
  liangel-02

  committed
  6e31683

  Copy full SHA for 6e31683

• Update on "add int4tensor support for safetensors"

  Show description for 60c5de9
  liangel-02

  committed
  60c5de9

  Copy full SHA for 60c5de9

• Update on "add int4tensor support for safetensors"

  Show description for 0200ad1
  liangel-02

  committed
  0200ad1

  Copy full SHA for 0200ad1

• Update on "add int4tensor support for safetensors"

  Show description for bbeaf52
  liangel-02

  committed
  bbeaf52

  Copy full SHA for bbeaf52

• Update on "add int4tensor support for safetensors"

  Show description for 7c82d4d
  liangel-02

  committed
  7c82d4d

  Copy full SHA for 7c82d4d

• add int4tensor support for safetensors

  Show description for e212a84
  liangel-02

  committed
  e212a84

  Copy full SHA for e212a84

• Add conversion options

  Show description for b47f1a3
  metascroy

  authored
  b47f1a3

  Copy full SHA for b47f1a3

• Revert "Unify get_block_size" (#3054)

  Show description for f92b898
  jerryzh168

  authored
  f92b898

  Copy full SHA for f92b898

Commits on Sep 23, 2025

• Unify get_block_size (#3039)

  Show description for 8e2ca35
  Xia-Weiwen

  authored
  8e2ca35

  Copy full SHA for 8e2ca35

Commits on Sep 22, 2025

• Minor fix on TAO op to support lowering

  Show description for eadead5
  RandySheriff

  authored
  eadead5

  Copy full SHA for eadead5

• Misc fixes for release scripts to make it easier to use (#3036)

  Show description for be4203e
  jerryzh168

  authored
  be4203e

  Copy full SHA for be4203e

• Avoid normalization layers in HF's quantization_config (#3030)

  Show description for fb7c837
  lisjin

  authored
  fb7c837

  Copy full SHA for fb7c837

• Update deprecated parameters in Hugging Face library (#2982)

  Show description for bc72e1c
  namgyu-youn

  authored
  bc72e1c

  Copy full SHA for bc72e1c

• [mxfp8 moe training] use new 3d colwise quantization kernel (#3037)

  danielvegamyhre

  authored
  9d88c16

  Copy full SHA for 9d88c16

Commits on Sep 21, 2025

• [Float8] add non-decomposed version of quantize/dequantize ops for fp8 (#2961)

  Show description for 8525185
  shiyang-weng

  authored
  8525185

  Copy full SHA for 8525185

• [Bug fix][CPU] Fix fp8 sdpa compiling issue with latest PyTorch (#2991)

  Show description for 22819f4
  Valentine233

  authored
  22819f4

  Copy full SHA for 22819f4

Commits on Sep 20, 2025

• [mxfp8 moe training] update 3d quant colwise scaling kernel to use single input/output TMA descriptors (#3034)

  danielvegamyhre

  authored
  d2fae7a

  Copy full SHA for d2fae7a

Commits on Sep 19, 2025

• Pass QAT learned qparams in convert (#3022)

  Show description for ae12e42
  andrewor14

  authored
  ae12e42

  Copy full SHA for ae12e42

• [mxfp8 moe training] remove mxfp8_gemms.py (#3033)

  danielvegamyhre

  authored
  f35dcd7

  Copy full SHA for f35dcd7

• [mxfp8 moe training] wrap 3d quantize tensor in custom ops and integrate it (#3004)

  danielvegamyhre

  authored
  4bf39b0

  Copy full SHA for 4bf39b0

• [mxfp8 moe training] add CUDA kernel to quantize 3d tensor colwise (#3002)

  danielvegamyhre

  authored
  f210443

  Copy full SHA for f210443

• Support Int4OpaqueTensor for HQQ (#3028)

  Show description for 1591603
  cyxlily

  authored
  1591603

  Copy full SHA for 1591603

• Add main tensor conversion API for packed tensors (#3029)

  Show description for a951643
  jerryzh168

  authored
  a951643

  Copy full SHA for a951643

• Support PLAIN_INT32 for AWQ on Intel GPU (#3019)

  Show description for cfa39c8
  xiaowangintel

  authored
  cfa39c8

  Copy full SHA for cfa39c8

• Remove FbgemmConfig and remaining Fbgemm tensors (#3032)

  Show description for ae204cc
  jerryzh168

  authored
  ae204cc

  Copy full SHA for ae204cc

Commits on Sep 18, 2025

• update compile arg for llama3.sh bench script (#3006)

  danielvegamyhre

  authored
  18dbe87

  Copy full SHA for 18dbe87

• Fix torchao_convert, remove StretchedAffineQuantizedTensor (#3015)

  lisjin

  authored
  122b307

  Copy full SHA for 122b307

Commits on Sep 17, 2025

• [sparse] Add in missing op support for FP8 Sparse (#3014)

  Show description for c801f10
  jcaip

  authored
  c801f10

  Copy full SHA for c801f10

• [moe training] add benchmarks for dsv3 236b, 671b shapes; reorganize benchmarks dir (#2999)

  danielvegamyhre

  authored
  f75b251

  Copy full SHA for f75b251

• [mxfp8 moe training] use dim1 cast cuda kernel for 3d weights by reshaping to 2d (#2998)

  danielvegamyhre

  authored
  ff3ba31

  Copy full SHA for ff3ba31

Pagination

Previous

Next