Commits

Branch selector

 main

User selector

All users

Datepicker

All time

Commit History

Commits on Nov 24, 2021

• Avoid clang format when applying CUB prefix/suffix header file (pytorch#772)

  Show description for 6cd2bde
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  6cd2bde

• Remove CUDA deps from tbe_input_combine_with_length_cpu

  Show description for 6658e19
  

  houseroad

  authored and

  facebook-github-bot

  committed
  6658e19

Commits on Nov 23, 2021

• create TBE inputs combine ops for length (pytorch#770)

  Show description for 4e81528
  

  Ruichao Xiao

  authored and

  facebook-github-bot

  committed
  4e81528

• Tidy up pytorch/FBGEMM copyright headers. (pytorch#765)

  Show description for 64d01d0
  

  Rick Weyrauch

  authored and

  facebook-github-bot

  committed
  64d01d0

• Remove CUDA deps of input_combine_cpu (pytorch#768)

  Show description for 36e8219
  

  houseroad

  authored and

  facebook-github-bot

  committed
  36e8219

• Fix formatting of FBGEMM_GPU README. (pytorch#769)

  Show description for 832e6fe
  

  Rick Weyrauch

  authored and

  facebook-github-bot

  committed
  832e6fe

Commits on Nov 22, 2021

• Add env config with the new OSS build (pytorch#767)

  Show description for a4262db
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  a4262db

• Add missing optimizers to FBGEMM_GPU OSS build. (pytorch#766)

  Show description for b3e3b86
  

  Rick Weyrauch

  authored and

  facebook-github-bot

  committed
  b3e3b86

• skbuild/cmake setup.py for fbgemm_gpu (pytorch#665)

  Show description for 41ab971
  

  Stephan Uphoff

  authored and

  facebook-github-bot

  committed
  41ab971

Commits on Nov 20, 2021

• Add int64_t and float support to rebatching CPU ops. (pytorch#764)

  Show description for e1e800b
  

  Rick Weyrauch

  authored and

  facebook-github-bot

  committed
  e1e800b

Commits on Nov 19, 2021

• Implement exact rowwise weighted adagrad, API change

  Show description for f88aa0a
  

  Wang Zhou

  authored and

  facebook-github-bot

  committed
  f88aa0a

Commits on Nov 18, 2021

• Use output_dtype as the argument for determing the output type for training TBE (pytorch#762)

  Show description for 52af920
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  52af920

• Add FP32 weight support in quantization converter and disagg publisher (pytorch#763)

  Show description for 0b5d82b
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  0b5d82b

• Revert D31843250: open-source segment_sum_csr()

  Show description for 03d891d
  

  shintaro-iwasaki

  authored and

  facebook-github-bot

  committed
  03d891d

Commits on Nov 17, 2021

• support different dtype for test_reorder_batched_ad_indices

  Show description for 4551833
  

  Zhen Qin

  authored and

  facebook-github-bot

  committed
  4551833

• open-source segment_sum_csr() (pytorch#754)

  Show description for 858365f
  

  shintaro-iwasaki

  authored and

  facebook-github-bot

  committed
  858365f

Commits on Nov 16, 2021

• Add output_dtype in v0 version of training TBE op (pytorch#18)

  Show description for 7e1183c
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  7e1183c

Commits on Nov 12, 2021

• support different dtype for reorder_batched_ad_lengths_gpu (pytorch#755)

  Show description for f3d3b95
  

  Luoshang Pan

  authored and

  facebook-github-bot

  committed
  f3d3b95

• optimize csr2csc for non shared table (pytorch#757)

  Show description for 4399bed
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  4399bed

• use SOA in radix sort (pytorch#756)

  Show description for 8db978a
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  8db978a

• Add output_dtype support for infer TBE (pytorch#749)

  Show description for eb23a75
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  eb23a75

Commits on Nov 11, 2021

• opensource pad sequence kernels

  Show description for c217614
  

  xing-liu

  authored and

  facebook-github-bot

  committed
  c217614

• Revert D32088453: call v2 function from the op and then add output_dtype into v0 function

  Show description for 9985369
  

  xw285cornell

  authored and

  facebook-github-bot

  committed
  9985369

Commits on Nov 10, 2021

• call v2 function from the op and then add output_dtype into v0 function

  Show description for e555516
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  e555516

• fix bug in CPU exact rowwise adagrad for shared table (pytorch#753)

  Show description for 7eff3ae
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  7eff3ae

Commits on Nov 8, 2021

• Unit test robustness improvments. (pytorch#752)

  Show description for 1e392b5
  

  Rick Weyrauch

  authored and

  facebook-github-bot

  committed
  1e392b5

Commits on Nov 5, 2021

• throw an exception when use 2-bit quantization (pytorch#751)

  Show description for 58959dd
  

  Cheng Cai

  authored and

  facebook-github-bot

  committed
  58959dd

Commits on Nov 4, 2021

• Replace #include "autograd/custom_function.h" which breaks build

  Show description for f185ce8
  

  samiwilf

  authored and

  facebook-github-bot

  committed
  f185ce8

Commits on Nov 3, 2021

• Move batched_unary_embeddings from hpc/ops to fbgemm_gpu. (pytorch#748)

  Show description for 5e2bf5d
  

  Rick Weyrauch

  authored and

  facebook-github-bot

  committed
  5e2bf5d

Commits on Nov 2, 2021

• support mix int32 and int64 of TBE input (pytorch#747)

  Show description for e2749d1
  

  Ruichao Xiao

  authored and

  facebook-github-bot

  committed
  e2749d1

Commits on Nov 1, 2021

• Try v2 solution

  Show description for a0c9ea3
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  a0c9ea3

• Attempt to resolve intermittant long-running (time-out) unit tests.

  Show description for 15fb88f
  

  Rick Weyrauch

  authored and

  facebook-github-bot

  committed
  15fb88f

Commits on Oct 29, 2021

• Reduce run-times of fbgemm_gpu unit tests.

  Show description for 2e15f56
  

  Rick Weyrauch

  authored and

  facebook-github-bot

  committed
  2e15f56

Commits on Oct 28, 2021

• TBE input combine (pytorch#743)

  Show description for 09d728e
  

  842974287

  authored and

  facebook-github-bot

  committed
  09d728e

Commits on Oct 27, 2021

• Implement exact rowwise weighted adagrad (pytorch#742)

  Show description for 37951bd
  

  Wang Zhou

  authored and

  facebook-github-bot

  committed
  37951bd

Pagination

PreviousNext