Commits

Branch selector

 f433d5f

User selector

All users

Datepicker

All time

Commit History

Commits on Mar 20, 2021

• format split_table_batched_embeddings_ops.py (pytorch#563)

  Show description for 2ab5e3f
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  2ab5e3f

Commits on Mar 19, 2021

• correctly handle fp16 table in approx sgd

  Show description for a8155a0
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  a8155a0

• Also include the dependency to FBGEMM CPU SPMDM routine (pytorch#559)

  Show description for e297346
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  e297346

Commits on Mar 18, 2021

• Optimize the performance of recat_embedding_grad_output (pytorch#488)

  Show description for d4f2dcc
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  d4f2dcc

• Add FBGEMM_CPU dependency in setup.py for table batched embedding cpu op (pytorch#556)

  Show description for e4ca7c4
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  e4ca7c4

• Fix the OSS CMakefile build (pytorch#555)

  Show description for 7716e15
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  7716e15

• only test optimizers supported in CPU in test_backward_optimizers (pytorch#557)

  Show description for a064fec
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  a064fec

• Remove "step" from table batched Rowwise Adagrad optimizer state (pytorch#481)

  Show description for 51bc2ec
  

  Wen Zhang (AI Infra)

  authored and

  facebook-github-bot

  committed
  51bc2ec

• Fix OSS build and run issues (pytorch#554)

  Show description for 4ef1ec0
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  4ef1ec0

• Add per-feature precision support in the model level (pytorch#552)

  Show description for 334134c
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  334134c

Commits on Mar 17, 2021

• real sparse fused exact optimizer: SGD and (row-wise) AdaGrad (pytorch#551)

  Show description for 95800eb
  

  jiyuanzFB

  authored and

  facebook-github-bot

  committed
  95800eb

• use fbgemm cpu embedding spmdm jit kernel (pytorch#550)

  Show description for a06cb38
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  a06cb38

• add output stride option to embedding spmdm (pytorch#548)

  Show description for 4ab1eeb
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  4ab1eeb

• Remove dead includes in deeplearning/fbgemm/test (pytorch#534)

  Show description for 88ba128
  

  generatedunixname89002005287564

  authored and

  facebook-github-bot

  committed
  88ba128

Commits on Mar 16, 2021

• Forcing grad_output to be 16 bytes aligned

  Show description for 8ae350d
  

  xing-liu

  authored and

  facebook-github-bot

  committed
  8ae350d

• use 2D parallelism in CPU emb fwd (pytorch#549)

  Show description for a4d3ca7
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  a4d3ca7

• Update asmjit for fbcode and oss (pytorch#538)

  Show description for 5d36b7d
  

  dskhudia

  authored and

  facebook-github-bot

  committed
  5d36b7d

• have bf16 quantize kernels use 2d blocks with loop to cover large tensors (pytorch#541)

  Show description for 2b5c02d
  

  amylittleyang

  authored and

  facebook-github-bot

  committed
  2b5c02d

• Register cpu path table_batched parameters as buffers to enable running on Mast (pytorch#547)

  Show description for bd470d9
  

  Chris Cai

  authored and

  facebook-github-bot

  committed
  bd470d9

Commits on Mar 15, 2021

• use at::acc_type (pytorch#546)

  Show description for 4e7dbd0
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  4e7dbd0

• parallelize across T rather than B to avoid race condition in exact sgd (pytorch#544)

  Show description for 316ef4f
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  316ef4f

Commits on Mar 13, 2021

• Optimizing cache memory (pytorch#543)

  Show description for a7fd8fb
  

  Junjie Yang

  authored and

  facebook-github-bot

  committed
  a7fd8fb

Commits on Mar 11, 2021

• Enable enforce_hbm option to allow placing both emb and cache in HBM (pytorch#539)

  Show description for 81738bb
  

  amylittleyang

  authored and

  facebook-github-bot

  committed
  81738bb

Commits on Mar 10, 2021

• print total execution time of mkl (pytorch#537)

  Show description for d12fc48
  

  YangWang92

  authored and

  facebook-github-bot

  committed
  d12fc48

Commits on Mar 9, 2021

• Multi-threading support for table batched embedding ops (cpu mode) (pytorch#535)

  Show description for 06f5f62
  

  mustafaozdal

  authored and

  facebook-github-bot

  committed
  06f5f62

• Add the cmake file for searching sphinx (pytorch#536)

  Show description for 4b88f40
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  4b88f40

Commits on Mar 8, 2021

• Clang Format (pytorch#533)

  Show description for 25d1620
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  25d1620

• Performance improvements for embedding index weight gradient calculation (cpu mode)

  Show description for da1e687
  

  mustafaozdal

  authored and

  facebook-github-bot

  committed
  da1e687

Commits on Mar 7, 2021

• Enabling codegen for approximate sparse optimizers for embedding bag

  Show description for becaf15
  

  mustafaozdal

  authored and

  facebook-github-bot

  committed
  becaf15

Commits on Mar 5, 2021

• hfp8 conversion kernels (pytorch#531)

  Show description for 53c7846
  

  jiyuanzFB

  authored and

  facebook-github-bot

  committed
  53c7846

• Remember cuda device for managed allocations and use it to release the managed memory. (pytorch#530)

  Show description for 6ee23c2
  

  Stephan Uphoff

  authored and

  facebook-github-bot

  committed
  6ee23c2

• Improving performance of table batched embedding backward for cpu mode

  Show description for 186377a
  

  mustafaozdal

  authored and

  facebook-github-bot

  committed
  186377a

Commits on Mar 4, 2021

• support 8bit rowwise quantize op in OSS tests (pytorch#522)

  Show description for 7dd5547
  

  amylittleyang

  authored and

  facebook-github-bot

  committed
  7dd5547

Commits on Mar 3, 2021

• Add Copyright Headers to comply the OSS requirements (pytorch#529)

  Show description for e0a2e71
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  e0a2e71

• fix split_table_batched_embeddings_test on cpu (pytorch#524)

  Show description for 2b53662
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  2b53662

Pagination

PreviousNext