Commits

Breadcrumbs

History for

onmaster

User selector

All users

Datepicker

All time

Commit History

Commits on Mar 2, 2020

• Fix the errror in logging SpMM generated code (pytorch#312)

  Show description for af57f36
  

  jiecaoyu

  authored and

  facebook-github-bot

  committed
  af57f36

  Copy full SHA for af57f36

Commits on Mar 1, 2020

• Fix the error in calling getCodeLoggingFile() (pytorch#311)

  Show description for baab538
  

  jiecaoyu

  authored and

  facebook-github-bot

  committed
  baab538

  Copy full SHA for baab538

Commits on Feb 27, 2020

• fbgemm handles block size 1 sparse adagrad (pytorch#306)

  Show description for 967d4bc
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  967d4bc

  Copy full SHA for 967d4bc

Commits on Feb 24, 2020

• enable parallel for quantization (pytorch#305)

  Show description for e1b1a55
  

  Hongzhang Shan

  authored and

  facebook-github-bot

  committed
  e1b1a55

  Copy full SHA for e1b1a55

Commits on Feb 19, 2020

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

  Show description for d9ead2d
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  d9ead2d

  Copy full SHA for d9ead2d

• clean up embedding spmdm tests (pytorch#300)

  Show description for f40e6d1
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  f40e6d1

  Copy full SHA for f40e6d1

Commits on Feb 17, 2020

• make order btw div and mul in adgrad consistent (#32974)

  Show description for 19c040c
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  19c040c

  Copy full SHA for 19c040c

• rowwise sparse adagrad fused (pytorch#298)

  Show description for b2fceb9
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  b2fceb9

  Copy full SHA for b2fceb9

Commits on Feb 15, 2020

• simplify avx2 mask usage in embedding kernels (pytorch#287)

  Show description for 618f71a
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  618f71a

  Copy full SHA for 618f71a

Commits on Feb 12, 2020

• export more classes for pytorch (pytorch#295)

  Show description for 40d6909
  

  Hongzhang Shan

  authored and

  facebook-github-bot

  committed
  40d6909

  Copy full SHA for 40d6909

Commits on Feb 10, 2020

• Reuse the partition scheme for GEMM in Depthwise Convolution (pytorch#293)

  Show description for 07dbb5d
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  07dbb5d

  Copy full SHA for 07dbb5d

• More fixes to enable USE_FBGEMM for PyTorch on Windows (pytorch#291)

  Show description for dcde869
  

  Hongzhang Shan

  authored and

  facebook-github-bot

  committed
  dcde869

  Copy full SHA for dcde869

Commits on Feb 7, 2020

• clean up temporary code for isa-dependent serialized packed weight (pytorch#290)

  Show description for 880ade1
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  880ade1

  Copy full SHA for 880ade1

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

  Show description for 8a21a3b
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  8a21a3b

  Copy full SHA for 8a21a3b

Commits on Feb 4, 2020

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

  Show description for e77369b
  

  Hongzhang Shan

  authored and

  facebook-github-bot

  committed
  e77369b

  Copy full SHA for e77369b

• enable positional weight for EmbeddingSpMDMNBitRowWiseSparse (pytorch#281)

  Show description for 5cdd1ab
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  5cdd1ab

  Copy full SHA for 5cdd1ab

• enable EmbeddingSpMDMNBitRowWiseSparse avx2 (pytorch#280)

  Show description for 07fba56
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  07fba56

  Copy full SHA for 07fba56

Commits on Feb 1, 2020

• add 2-stage SparseAdagrad interface (pytorch#276)

  Show description for 930ea23
  

  jiecaoyu

  authored and

  facebook-github-bot

  committed
  930ea23

  Copy full SHA for 930ea23

• clang-format the code (pytorch#278)

  Show description for fdb82b2
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  fdb82b2

  Copy full SHA for fdb82b2

• Fix the root cause for _aligned_free issues on row_offset_ allocation (pytorch#277)

  Show description for e4122f7
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  e4122f7

  Copy full SHA for e4122f7

Commits on Jan 31, 2020

• add more instantiation for EmbeddingSpMDMAvx2 (pytorch#274)

  Show description for a62fabf
  

  Hongzhang Shan

  authored and

  facebook-github-bot

  committed
  a62fabf

  Copy full SHA for a62fabf

• 2/4-bit EmbeddingSPMDM with rowwise sparsity (pytorch#263)

  Show description for 62c8286
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  62c8286

  Copy full SHA for 62c8286

• Enable FMA operation for windows (pytorch#273)

  Show description for edc4a4f
  

  Hongzhang Shan

  authored and

  facebook-github-bot

  committed
  edc4a4f

  Copy full SHA for edc4a4f

• 2bit JIT'ed SLS kernel (pytorch#234)

  Show description for 0e31e0a
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  0e31e0a

  Copy full SHA for 0e31e0a

Commits on Jan 29, 2020

• Add instantiation for template parameters

  Show description for b81d065
  

  Hongzhang Shan

  authored and

  facebook-github-bot

  committed
  b81d065

  Copy full SHA for b81d065

Commits on Jan 27, 2020

• make gconv work for any W and H (pytorch#269)

  Show description for ef4d11b
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  ef4d11b

  Copy full SHA for ef4d11b

Commits on Jan 22, 2020

• add missing copyright header (pytorch#267)

  Show description for 27cb280
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  27cb280

  Copy full SHA for 27cb280

• EmbeddingSpMDM with float16 input (pytorch#265)

  Show description for a1543b1
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  a1543b1

  Copy full SHA for a1543b1

• modified instrinsic fp16 kernel for windows build (pytorch#259)

  Show description for bd7becc
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  bd7becc

  Copy full SHA for bd7becc

• Intrinsic implementation of FP16 kernels for windows build (pytorch#254)

  Show description for 2a8eb50
  

  ykim362

  authored and

  facebook-github-bot

  committed
  2a8eb50

  Copy full SHA for 2a8eb50

Commits on Jan 21, 2020

• don't need extra register for positional weights (pytorch#261)

  Show description for 2df47f5
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  2df47f5

  Copy full SHA for 2df47f5

• specialize fp32 SLS for emb dim 1 (pytorch#264)

  Show description for a37cc62
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  a37cc62

  Copy full SHA for a37cc62

• avx2 4-bit SLS (pytorch#262)

  Show description for 2957bd4
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  2957bd4

  Copy full SHA for 2957bd4

• remove EmbeddingSpMDM4BitKernelSignature (pytorch#257)

  Show description for ab4da8f
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  ab4da8f

  Copy full SHA for ab4da8f

Commits on Jan 19, 2020

• Add github action for windows

  Show description for a22e6b8
  

  Hongzhang Shan

  authored and

  facebook-github-bot

  committed
  a22e6b8

  Copy full SHA for a22e6b8

Pagination

Previous

Next