Commits

Branch selector

 main

User selector

All users

Datepicker

All time

Commit History

Commits on Oct 12, 2021

• optimized cat open source (pytorch#654)

  Show description for 1cc3b2d
  

  YazhiGao

  authored and

  facebook-github-bot

  committed
  1cc3b2d

Commits on Oct 11, 2021

• pooled emb int8 quantization fusion: function def changes

  Show description for 6588f7c
  

  amylittleyang

  authored and

  facebook-github-bot

  committed
  6588f7c

• pooled emb int8 quantization fusion: kernel changes

  Show description for f6c9c2e
  

  amylittleyang

  authored and

  facebook-github-bot

  committed
  f6c9c2e

• fbgemm_gpu compiler flags (pytorch#724)

  Show description for 298dfc4
  

  r-barnes

  authored and

  facebook-github-bot

  committed
  298dfc4

Commits on Oct 9, 2021

• Change the input of merge_pooled_embeddings to be more specific inputs (pytorch#725)

  Show description for 127ebd3
  

  Mengchi Zhang

  authored and

  facebook-github-bot

  committed
  127ebd3

Commits on Oct 8, 2021

• ConvUnifiedBenchmark: Add new cmdline options (pytorch#723)

  Show description for 743a787
  

  digantdesai

  authored and

  facebook-github-bot

  committed
  743a787

• Move reorder_batched_ad_lengths and indices from caffe2/fb to fbgemm_gpu. (pytorch#722)

  Show description for bf5db18
  

  Rick Weyrauch

  authored and

  facebook-github-bot

  committed
  bf5db18

• Support inplace materialization for TBE (pytorch#726)

  Show description for 24d4233
  

  houseroad

  authored and

  facebook-github-bot

  committed
  24d4233

Commits on Oct 7, 2021

• SplitTableBatchedEmbeddings module changes for int8 quantize fusion

  Show description for 4b67768
  

  amylittleyang

  authored and

  facebook-github-bot

  committed
  4b67768

• OSS Permute_pooled_embeddings_ops (pytorch#717)

  Show description for 6a44e7f
  

  RenfeiChen-FB

  authored and

  facebook-github-bot

  committed
  6a44e7f

• Remove unused input from inference TBE (pytorch#720)

  Show description for 3225240
  

  houseroad

  authored and

  facebook-github-bot

  committed
  3225240

Commits on Oct 5, 2021

• async _to_offsets(...) for KJT

  Show description for 5243fc4
  

  xing-liu

  authored and

  facebook-github-bot

  committed
  5243fc4

Commits on Sep 30, 2021

• Add a warning message for CUB DIR and cuda build vs. non-cuda build in setup.py

  Show description for 96e7cd8
  

  Rick Weyrauch

  authored and

  facebook-github-bot

  committed
  96e7cd8

Commits on Sep 28, 2021

• Enable indices bounds check by default

  Show description for 7bee788
  

  Andrew Tulloch

  authored and

  facebook-github-bot

  committed
  7bee788

• Add pooled embedding permute (pytorch#714)

  Show description for 581277e
  

  amylittleyang

  authored and

  facebook-github-bot

  committed
  581277e

Commits on Sep 27, 2021

• Remove numpy dependency in split_table_batched_embeddings_ops.py

  Show description for 38fbf43
  

  Rick Weyrauch

  authored and

  facebook-github-bot

  committed
  38fbf43

Commits on Sep 24, 2021

• initialize op weights in hpc publisher before loading weights from checkpoint (pytorch#712)

  Show description for 2728266
  

  amylittleyang

  authored and

  facebook-github-bot

  committed
  2728266

Commits on Sep 23, 2021

• Split up long running tests to avoid timeouts when running tests are part of the CI pipeline. (pytorch#711)

  Show description for 7588d9d
  

  Rick Weyrauch

  authored and

  facebook-github-bot

  committed
  7588d9d

• Fix CPU-only build for both Buck and OSS builds. (pytorch#710)

  Show description for c1b3d99
  

  Rick Weyrauch

  authored and

  facebook-github-bot

  committed
  c1b3d99

Commits on Sep 22, 2021

• remove weight initialization in table batched emb init (pytorch#709)

  Show description for 10c3f6a
  

  amylittleyang

  authored and

  facebook-github-bot

  committed
  10c3f6a

• shuffle indices generated by zipf to avoid unintended spatial locality (pytorch#708)

  Show description for b119fc6
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  b119fc6

• Add FP32 support in infer version of table batched embedding op (pytorch#707)

  Show description for 87f4f4f
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  87f4f4f

Commits on Sep 21, 2021

• Reduce maximum number of test iterations to reduce likelyhood of test timeouts. (pytorch#706)

  Show description for 845aece
  

  Rick Weyrauch

  authored and

  facebook-github-bot

  committed
  845aece

Commits on Sep 20, 2021

• Fix some fbgemm_gpu types (pytorch#704)

  Show description for 97c45d9
  

  r-barnes

  authored and

  facebook-github-bot

  committed
  97c45d9

• Fix OSS build errors with Tesla T4 using sm75 compute (pytorch#703)

  Show description for 0108d4f
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  0108d4f

Commits on Sep 18, 2021

• suppress errors in deeplearning/fbgemm/fbgemm_gpu

  Show description for 2351fe3
  

  Pyre Bot Jr

  authored and

  facebook-github-bot

  committed
  2351fe3

Commits on Sep 17, 2021

• Manually cache block attributes in PackBMatrix::pack_unpack_ (pytorch#702)

  Show description for dccc573
  

  swolchok

  authored and

  facebook-github-bot

  committed
  dccc573

• Improve load balancing of long segments across blocks

  Show description for 321db4d
  

  Andrew Tulloch

  authored and

  facebook-github-bot

  committed
  321db4d

Commits on Sep 16, 2021

• remove pendatic to avoid compile error (pytorch#701)

  Show description for 91fc122
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  91fc122

Commits on Sep 15, 2021

• Add pruning support into disagg publisher (pytorch#700)

  Show description for e187e74
  

  ipiszy

  authored and

  facebook-github-bot

  committed
  e187e74

Commits on Sep 14, 2021

• update README

  Show description for 4d9c006
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  4d9c006

• Implement a separate kernel for array-based index remapping lookup. (pytorch#682)

  Show description for 820b4c7
  

  ipiszy

  authored and

  facebook-github-bot

  committed
  820b4c7

Commits on Sep 13, 2021

• Support weights_uvm on CPU ops

  Show description for 7ceee02
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  7ceee02

• make the CPU and CUDA op interface of infer version of table batched embedding bag the same (pytorch#696)

  Show description for 5d100bd
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  5d100bd

• Rename ad_pooled_embeddings to pooled_embeddings (pytorch#695)

  Show description for 9c12ad2
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  9c12ad2

Pagination

PreviousNext