Commits

Breadcrumbs

History for

onF4

User selector

All users

Datepicker

All time

Commit History

Commits on Aug 7, 2020

• enable 3x5x5 and 5x5x5 3d depthwise conv (pytorch#399)

  Show description for 0b01501
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  0b01501

  Copy full SHA for 0b01501

Commits on Jul 21, 2020

• Bug fix in depthwise conv (pytorch#395)

  Show description for 139c6f2
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  139c6f2

  Copy full SHA for 139c6f2

Commits on Jul 3, 2020

• avx2 fused quant dequant kernel (pytorch#389)

  Show description for 503e033
  

  amylittleyang

  authored and

  facebook-github-bot

  committed
  503e033

  Copy full SHA for 503e033

Commits on Jun 20, 2020

• add more comments and make test name more accurate (pytorch#388)

  Show description for 9b81c46
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  9b81c46

  Copy full SHA for 9b81c46

Commits on Jun 1, 2020

• minor fix (pytorch#385)

  Show description for 15a6c90
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  15a6c90

  Copy full SHA for 15a6c90

Commits on May 16, 2020

• Float16 rowwise sparse adagrad with stochastic rounding (reattempt of D21519194) (pytorch#375)

  Show description for c881ebf
  

  uyongw

  authored and

  facebook-github-bot

  committed
  c881ebf

  Copy full SHA for c881ebf

• Back out "Float16 rowwise sparse adagrad with stochastic rounding"

  Show description for 9f5ac27
  

  Natalia Gimelshein

  authored and

  facebook-github-bot

  committed
  9f5ac27

  Copy full SHA for 9f5ac27

Commits on May 15, 2020

• Float16 rowwise sparse adagrad with stochastic rounding (pytorch#370)

  Show description for b9f1a80
  

  uyongw

  authored and

  facebook-github-bot

  committed
  b9f1a80

  Copy full SHA for b9f1a80

• Add conv_1d (pytorch#369)

  Show description for 17b31be
  

  Hongzhang Shan

  authored and

  facebook-github-bot

  committed
  17b31be

  Copy full SHA for 17b31be

Commits on May 14, 2020

• dirsync fbgemm with xplat

  Show description for 7ed5f9f
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  7ed5f9f

  Copy full SHA for 7ed5f9f

Commits on May 12, 2020

• minor fix (pytorch#371)

  Show description for 7bb3cd7
  

  Hongzhang Shan

  authored and

  facebook-github-bot

  committed
  7bb3cd7

  Copy full SHA for 7bb3cd7

Commits on May 6, 2020

• unified conv to call dw conv with 2 oc per g (pytorch#360)

  Show description for 40f530d
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  40f530d

  Copy full SHA for 40f530d

• depthwise convolution with 2 output channels per group (pytorch#359)

  Show description for dfbb13f
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  dfbb13f

  Copy full SHA for dfbb13f

Commits on May 1, 2020

• remove one-stage interface for sparse adagrad (pytorch#366)

  Show description for b59c633
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  b59c633

  Copy full SHA for b59c633

Commits on Apr 30, 2020

• implement L2 regularization for Adagrad in fbgemm (pytorch#365)

  Show description for 17d9a60
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  17d9a60

  Copy full SHA for 17d9a60

Commits on Apr 13, 2020

• minor changes in UniConvTest and GConvTest (pytorch#357)

  Show description for 15e343c
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  15e343c

  Copy full SHA for 15e343c

• follow EXPECT_EQ(actual, expected) convention (pytorch#354)

  Show description for f47baf3
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  f47baf3

  Copy full SHA for f47baf3

Commits on Apr 10, 2020

• add option to use int64_t offsets/lengths in embedding operators (pytorch#350)

  Show description for 3db8b84
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  3db8b84

  Copy full SHA for 3db8b84

• Move FakeFP16 back to internal to remove dependency on MKL (#36297)

  Show description for 20b6cf1
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  20b6cf1

  Copy full SHA for 20b6cf1

Commits on Apr 9, 2020

• Open sourcing fbgemm_fp16 ops (#36212)

  Show description for 1910f8c
  

  Yinghai Lu

  authored and

  facebook-github-bot

  committed
  1910f8c

  Copy full SHA for 1910f8c

Commits on Mar 31, 2020

• add offset-based interface for PyTorch (pytorch#334)

  Show description for df34258
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  df34258

  Copy full SHA for df34258

Commits on Mar 21, 2020

• Quantize should work even if the value after quantization becomes inf (pytorch#317)

  Show description for 3ee0b7d
  

  dskhudia

  authored and

  jspark1105

  committed
  3ee0b7d

  Copy full SHA for 3ee0b7d

• Revert D16234008: Quantize should work even if the value after quantization becomes inf

  Show description for ab431ad
  

  ezyang

  authored and

  jspark1105

  committed
  ab431ad

  Copy full SHA for ab431ad

• Quantize should work even if the value after quantization becomes inf (pytorch#107)

  Show description for 2a50f19
  

  dskhudia

  authored and

  jspark1105

  committed
  2a50f19

  Copy full SHA for 2a50f19

• Make spmmfp32 and spmmi8 benchmark consistent with tests (pytorch#315)

  Show description for 40235b3
  

  dskhudia

  authored and

  jspark1105

  committed
  40235b3

  Copy full SHA for 40235b3

• reduce Windows CI warnings (pytorch#309)

  Show description for 99df751
  jspark1105

  committed
  99df751

  Copy full SHA for 99df751

• make cmake less verbose (pytorch#308)

  Show description for 4942ea9
  jspark1105

  committed
  4942ea9

  Copy full SHA for 4942ea9

• add copyright to EmbeddingSpMDMTestUtils.{cc,h} (pytorch#303)

  Show description for ca1e41c
  jspark1105

  committed
  ca1e41c

  Copy full SHA for ca1e41c

• fix index mapping table type to int32_t (pytorch#302)

  Show description for 1a7be35
  jspark1105

  committed
  1a7be35

  Copy full SHA for 1a7be35

• clean up embedding spmdm tests (pytorch#300)

  Show description for 6a4ed1c
  jspark1105

  committed
  6a4ed1c

  Copy full SHA for 6a4ed1c

• rowwise sparse adagrad fused (pytorch#298)

  Show description for cfda552
  jspark1105

  committed
  cfda552

  Copy full SHA for cfda552

• Add ASAN check for github action (pytorch#294)

  Show description for 878e132
  

  Hongzhang Shan

  authored and

  jspark1105

  committed
  878e132

  Copy full SHA for 878e132

• row-wise sparse fp32/fp16/int8 EmbeddingSpMDM (pytorch#288)

  Show description for b8297c2
  jspark1105

  committed
  b8297c2

  Copy full SHA for b8297c2

• changes for shared release on windows (pytorch#284)

  Show description for d1ebf4e
  

  Hongzhang Shan

  authored and

  jspark1105

  committed
  d1ebf4e

  Copy full SHA for d1ebf4e

• enable positional weight for EmbeddingSpMDMNBitRowWiseSparse (pytorch#281)

  Show description for fb7a373
  jspark1105

  committed
  fb7a373

  Copy full SHA for fb7a373

Pagination

Previous

Next