Commits

Breadcrumbs

History for

onmain

User selector

All users

Datepicker

All time

Commit History

Commits on Feb 12, 2022

• Microbenchmarking for lxu_cache_lookup.

  Show description for ad7ae91
  

  jasonjk-park

  authored and

  facebook-github-bot

  committed
  ad7ae91

Commits on Feb 11, 2022

• Remove cache -> weights update as inference does not require it. (pytorch#928)

  Show description for b2226d2
  

  jasonjk-park

  authored and

  facebook-github-bot

  committed
  b2226d2

• clean up unused variables in int_nbit_split_embedding_codegen_forward_cpu (pytorch#927)

  Show description for 19801c1
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  19801c1

Commits on Feb 10, 2022

• Optimize linearize_cache_indices implementation. (pytorch#925)

  Show description for 806addf
  

  jasonjk-park

  authored and

  facebook-github-bot

  committed
  806addf

• Add INT8 and INT4 support to P2P benchmark. (pytorch#918)

  Show description for 501dfa7
  

  caogao

  authored and

  facebook-github-bot

  committed
  501dfa7

• Add tests to verify linearize_cache_indices. (pytorch#924)

  Show description for e385d02
  

  jasonjk-park

  authored and

  facebook-github-bot

  committed
  e385d02

• Microbenchmark for linearize_cache_indices (pytorch#926)

  Show description for a358709
  

  jasonjk-park

  authored and

  facebook-github-bot

  committed
  a358709

Commits on Feb 9, 2022

• Fix TorchRec cpu only unit tests (pytorch#921)

  Show description for 08ffd63
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  08ffd63

• Fix TorchRec CPU only build (pytorch#920)

  Show description for fa2a8ec
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  fa2a8ec

Commits on Feb 8, 2022

• Update README (pytorch#917)

  Show description for 969941d
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  969941d

• Use const& to avoid unnecessary copy. (pytorch#911)

  Show description for a613cc8
  

  jasonjk-park

  authored and

  facebook-github-bot

  committed
  a613cc8

• Continued migration of ops from the fb to fbgemm namespace. (pytorch#885)

  Show description for 685f878
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  685f878

• Add nighly build support (pytorch#912)

  Show description for 292e103
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  292e103

Commits on Feb 5, 2022

• Avoid unnecessary copy if possible. (pytorch#902)

  Show description for ddb8d25
  

  jasonjk-park

  authored and

  facebook-github-bot

  committed
  ddb8d25

• use thread local code cache in cpu quantized emb fwd (pytorch#903)

  Show description for e779169
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  e779169

Commits on Feb 3, 2022

• Fix unreferenced var error in quantize_ops.cu (pytorch#898)

  Show description for 66a9537
  

  r-barnes

  authored and

  facebook-github-bot

  committed
  66a9537

Commits on Feb 2, 2022

• Add annotations to deeplearning/fbgemm/fbgemm_gpu

  Show description for 2f935b9
  

  Pyre Bot Jr

  authored and

  facebook-github-bot

  committed
  2f935b9

• fix a bug in 2 cards setup (pytorch#894)

  Show description for 49fe829
  

  842974287

  authored and

  facebook-github-bot

  committed
  49fe829

Commits on Feb 1, 2022

• Integrate zero-ing into kernel for performance. (pytorch#890)

  Show description for 1280f81
  

  jasonjk-park

  authored and

  facebook-github-bot

  committed
  1280f81

Commits on Jan 31, 2022

• reduce compilation size and time (pytorch#886)

  Show description for bf48eea
  

  shintaro-iwasaki

  authored and

  facebook-github-bot

  committed
  bf48eea

• Set TORCH_CUDA_ARCH_LIST as command line arg rather than env var. (pytorch#889)

  Show description for 82c5acd
  

  Rick Weyrauch

  authored and

  facebook-github-bot

  committed
  82c5acd

Commits on Jan 27, 2022

• Update OSS README build instructions.

  Show description for 03a04eb
  

  Rick Weyrauch

  authored and

  facebook-github-bot

  committed
  03a04eb

Commits on Jan 25, 2022

• Bug fix when segment lengths are larger than 1. (pytorch#883)

  Show description for 4a1fdf5
  

  jasonjk-park

  authored and

  facebook-github-bot

  committed
  4a1fdf5

• remove grad_t to reduce binary size + compilation time (pytorch#882)

  Show description for f5bff95
  

  shintaro-iwasaki

  authored and

  facebook-github-bot

  committed
  f5bff95

• Bug fix for HBC by feature CPU implementation (pytorch#881)

  Show description for d41b8aa
  

  jasonjk-park

  authored and

  facebook-github-bot

  committed
  d41b8aa

• OSS fbgemm_gpu: Try to automatically find correct cuda installation in setup.py (pytorch#879)

  Show description for 3bc185b
  

  Stephan Uphoff

  authored and

  facebook-github-bot

  committed
  3bc185b

Commits on Jan 24, 2022

• use self.Assert in split_table_batched_embeddings_test (pytorch#880)

  Show description for ba72edd
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  ba72edd

Commits on Jan 21, 2022

• Add support for different types for segment value vs. length. (pytorch#878)

  Show description for 0406bf6
  

  jasonjk-park

  authored and

  facebook-github-bot

  committed
  0406bf6

• Add FBGEMM_GPU unit tests to GHA CPU-only build. (pytorch#875)

  Show description for 1b6412a
  

  Rick Weyrauch

  authored and

  facebook-github-bot

  committed
  1b6412a

Commits on Jan 20, 2022

• fbgemm_gpu: Remove unused file (pytorch#877)

  Show description for bef5415
  

  Stephan Uphoff

  authored and

  facebook-github-bot

  committed
  bef5415

• reduce memory used for test_stacked_jagged_2d_to_dense take 2 (pytorch#876)

  Show description for 17d6a15
  

  jspark1105

  authored and

  facebook-github-bot

  committed
  17d6a15

• Back out "add bfloat16 gradient option to split_table_batched_embeddings_benchmark (2/2)"

  Show description for 8275eb6
  

  Sudarshan Raghunathan

  authored and

  facebook-github-bot

  committed
  8275eb6

• Support other integral types for segment values in HBC by feature (pytorch#874)

  Show description for 9e8820b
  

  jasonjk-park

  authored and

  facebook-github-bot

  committed
  9e8820b

Commits on Jan 19, 2022

• add bfloat16 gradient option to split_table_batched_embeddings_benchmark (2/2)

  Show description for 48515f9
  

  shintaro-iwasaki

  authored and

  facebook-github-bot

  committed
  48515f9

• Update OSS README for FBGEMM_GPU (pytorch#869)

  Show description for 41ebf9b
  

  jianyuh

  authored and

  facebook-github-bot

  committed
  41ebf9b

Pagination

PreviousNext