Commits

Branch selector

 f433d5f

User selector

All users

Datepicker

All time

Commit History

Commits on Jun 7, 2021

• index remapping benchmark (pytorch#615)

  Show description for f433d5f
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  f433d5f

• Vectorize index remapping for pruned embedding tables (pytorch#607)

  Show description for 76cf190
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  76cf190

• work around 8-bit imm error in pytorch_bazel_build CI (pytorch#620)

  Show description for b3f1af5
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  b3f1af5

Commits on Jun 6, 2021

• int2/int4/int8/fp16 cpu/gpu mixed-precision kernel support

  Show description for 744179a
  

  Andrew Tulloch

  authored and

  facebook-github-bot

  committed
  744179a

Commits on Jun 5, 2021

• Add functions for dequantizing rowwise-quantized int2/int4/int8 + fp16 scale/bias to fp16 embeddings. (pytorch#619)

  Show description for 77a4792
  

  caogao

  authored and

  facebook-github-bot

  committed
  77a4792

Commits on Jun 3, 2021

• Remove usage of if constexpr from D28620537. (pytorch#618)

  Show description for 5885ee1
  

  caogao

  authored and

  facebook-github-bot

  committed
  5885ee1

Commits on Jun 2, 2021

• Add functions for quantizing fp16 embeddings to rowwise quantized int2/int4/int8 + fp16 scale/bias. (pytorch#616)

  Show description for 9cb33bc
  

  caogao

  authored and

  facebook-github-bot

  committed
  9cb33bc

Commits on May 24, 2021

• remove unncessary context saving (pytorch#613)

  Show description for ffc2e1a
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  ffc2e1a

• make functions defined in quantize_ops.cuh inline (pytorch#612)

  Show description for dd3abd3
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  dd3abd3

Commits on May 20, 2021

• add permute function to KJT

  Show description for 5bbb70d
  

  xing-liu

  authored and

  facebook-github-bot

  committed
  5bbb70d

Commits on May 19, 2021

• rollback part of D28312997 (pytorch#611)

  Show description for 442fb1d
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  442fb1d

Commits on May 17, 2021

• Fix the int8 embedding bench/test build (pytorch#610)

  Show description for d4a6222
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  d4a6222

Commits on May 16, 2021

• Fix the int4 table batched embedding benchmark with mixed dim (pytorch#609)

  Show description for c48ba9b
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  c48ba9b

• Andrew's int4 table batched embedding implementation (pytorch#608)

  Show description for 0fe80ee
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  0fe80ee

Commits on May 12, 2021

• add missing copyright headers

  Show description for 4b8aaad
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  4b8aaad

Commits on May 11, 2021

• remove fbgemm_gpu_test wrappers no longer used (pytorch#605)

  Show description for 7794b29
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  7794b29

• int8 table batched embedding bag on cpu (pytorch#603)

  Show description for dc66331
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  dc66331

• move quantize ops to fbgemm_gpu (pytorch#606)

  Show description for bcba0cf
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  bcba0cf

Commits on May 10, 2021

• format sparse_ops.cu (pytorch#604)

  Show description for 566d74c
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  566d74c

Commits on May 6, 2021

• fix ref call in RowWiseSparseAdagradFused (pytorch#602)

  Show description for 580d637
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  580d637

• Fix ref call in GenerateEmbeddingSpMDMWithStrides (pytorch#601)

  Show description for 5acd437
  

  malfet

  authored and

  facebook-github-bot

  committed
  5acd437

Commits on May 5, 2021

• Optimize _get_8bit_qparam_cuda_kernel to use warp sync primitives for the min and max reduction

  Show description for 95946ae
  

  AndrewGaspar

  authored and

  facebook-github-bot

  committed
  95946ae

Commits on May 4, 2021

• reduce # of barriers in cpu emb fwd (pytorch#599)

  Show description for 61f1cd7
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  61f1cd7

Commits on May 3, 2021

• handle empty indices (pytorch#600)

  Show description for 530356e
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  530356e

• format split_table_batched_embeddings_test

  Show description for fa45c15
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  fa45c15

Commits on Apr 30, 2021

• add license (pytorch#598)

  Show description for 5ce0eed
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  5ce0eed

Commits on Apr 28, 2021

• Permute Indices to FBGEMM (pytorch#597)

  Show description for b48fb95
  

  dstaay-fb

  authored and

  facebook-github-bot

  committed
  b48fb95

Commits on Apr 26, 2021

• Support for sequence embeddings (Embedding list no bag) (pytorch#582)

  Show description for 95805fb
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  95805fb

Commits on Apr 22, 2021

• suppress errors in deeplearning/fbgemm/fbgemm_gpu

  Show description for 4ceb117
  

  Pyre Bot Jr

  authored and

  facebook-github-bot

  committed
  4ceb117

Commits on Apr 21, 2021

• fix uvm bandwidth reporting bug and add flush gpu cache support (pytorch#596)

  Show description for cc4b585
  

  gsethi523

  authored and

  facebook-github-bot

  committed
  cc4b585

Commits on Apr 20, 2021

• per sample weight is always fp32 (pytorch#595)

  Show description for e1abaec
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  e1abaec

Commits on Apr 19, 2021

• allow all uvm table placement in split_table_batched_embeddings_benchmark (pytorch#594)

  Show description for 698c2dc
  

  gsethi523

  authored and

  facebook-github-bot

  committed
  698c2dc

Commits on Apr 14, 2021

• Add annotations to functions in stl/lightning/utilities/ema_state.py

  Show description for 625201f
  

  Thomas Chen

  authored and

  facebook-github-bot

  committed
  625201f

• row-wise sequence embeddings

  Show description for 0b3f25e
  

  YazhiGao

  authored and

  facebook-github-bot

  committed
  0b3f25e

Commits on Apr 11, 2021

• use f14 map for csr2csc (pytorch#592)

  Show description for c609f3c
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  c609f3c

Pagination

PreviousNext