Commits

Breadcrumbs

History for

onmaster

User selector

All users

Datepicker

All time

Commit History

Commits on Mar 30, 2022

• Conv-Bias-ReLU fusion (NVIDIA#1332)

  Show description for 23cfb57
  

  3 peopleauthored
  23cfb57

  Copy full SHA for 23cfb57

Commits on Mar 24, 2022

• Add CUDA Focal Loss Implementation (NVIDIA#1337)

  Show description for 28f8539
  

  crcrpar

  and

  alpha0422

  authored
  28f8539

  Copy full SHA for 28f8539

Commits on Feb 27, 2022

• build fused grad accum w/ wgrad only if cuda>10 (NVIDIA#1312)

  crcrpar

  authored
  47c269b

  Copy full SHA for 47c269b

Commits on Feb 26, 2022

• [transformer] Fuse grad accumulation with wgrad (NVIDIA#1297)

  Show description for ddc0803
  

  crcrpar

  and

  erhoo82

  authored
  ddc0803

  Copy full SHA for ddc0803

Commits on Feb 10, 2022

• 8.6 requires CUDA 11.1 (NVIDIA#1289)

  crcrpar

  authored
  e1aa1fc

  Copy full SHA for e1aa1fc

Commits on Feb 1, 2022

• Add the permutation related support as the extension for asp lib. (NVIDIA#1194)

  Show description for 89edb81
  ChongyuNVIDIA

  authored
  89edb81

  Copy full SHA for 89edb81

Commits on Jan 19, 2022

• pass flags to transducer joint kernel (NVIDIA#1273)

  crcrpar

  authored
  c4e85f7

  Copy full SHA for c4e85f7

Commits on Jan 13, 2022

• support new path to CUDAGeneratorImpl.h (NVIDIA#1267)

  shintaro-iwasaki

  authored
  b2fdf9c

  Copy full SHA for b2fdf9c

Commits on Dec 16, 2021

• version guard (NVIDIA#1253)

  crcrpar

  authored
  e847382

  Copy full SHA for e847382

Commits on Dec 15, 2021

• Add --threads 4 to extra_compile_args["nvcc"] (NVIDIA#1251)

  Show description for f63dac8
  crcrpar

  authored
  f63dac8

  Copy full SHA for f63dac8

Commits on Dec 14, 2021

• Faster --fast_multihead_attn build (NVIDIA#1245)

  Show description for 7ec8ed6
  crcrpar

  authored
  7ec8ed6

  Copy full SHA for 7ec8ed6

Commits on Dec 9, 2021

• Add fused mixed precision lamb optimizer. (NVIDIA#1237)

  Show description for 3c8f516
  kevinstephano

  authored
  3c8f516

  Copy full SHA for 3c8f516

Commits on Oct 27, 2021

• FastLayerNorm compat with autocast (NVIDIA#1203)

  Show description for ae75763
  

  crcrpar

  and

  yjk21

  authored
  ae75763

  Copy full SHA for ae75763

Commits on Oct 2, 2021

• transformer utils (NVIDIA#1181)

  Show description for 365fdc1
  

  5 peopleauthored
  365fdc1

  Copy full SHA for 365fdc1

Commits on Sep 8, 2021

• enable ninja (NVIDIA#1164)

  Show description for 9ce0a10
  crcrpar

  authored
  9ce0a10

  Copy full SHA for 9ce0a10

Commits on Sep 1, 2021

• Seryilmaz/fuse norm into scale (NVIDIA#1149)

  Show description for 4d190db
  

  seryilmaz

  and

  Sukru Eryilmaz

  authored
  4d190db

  Copy full SHA for 4d190db

• Seryilmaz/more cublas lt (NVIDIA#1147)

  Show description for 6af09dd
  

  seryilmaz

  and

  Sukru Eryilmaz

  authored
  6af09dd

  Copy full SHA for 6af09dd

Commits on Jul 17, 2021

• Added more fusion and vectorized kernel for transducer (NVIDIA#1125)

  Show description for 0c2c6ee
  nanz-nv

  authored
  0c2c6ee

  Copy full SHA for 0c2c6ee

• Adds small-batch kernels (NVIDIA#1126)

  yjk21

  authored
  ed71996

  Copy full SHA for ed71996

Commits on Apr 17, 2021

• Adding fast bottleneck implementation into contrib (NVIDIA#1079)

  Show description for 705cba9
  

  FDecaYed

  and

  ptrblck

  authored
  705cba9

  Copy full SHA for 705cba9

Commits on Apr 16, 2021

• adds fmhalib (NVIDIA#1074)

  yjk21

  authored
  5c9b21d

  Copy full SHA for 5c9b21d

Commits on Mar 24, 2021

• Initial check-in of the transducer extensions (NVIDIA#1069)

  Show description for d86d1b0
  nanz-nv

  authored
  d86d1b0

  Copy full SHA for d86d1b0

Commits on Feb 23, 2021

• fast layer norm (NVIDIA#1037)

  yjk21

  authored
  e2083df

  Copy full SHA for e2083df

Commits on Dec 1, 2020

• DistributedFusedAdam Model Parallelism Support (Megatron) (NVIDIA#981)

  Show description for 6b7e77b
  

  3 peopleauthored
  6b7e77b

  Copy full SHA for 6b7e77b

Commits on Aug 10, 2020

• move sm80 code inside MHA (NVIDIA#937)

  Show description for 5d9b5cb
  ptrblck

  authored
  5d9b5cb

  Copy full SHA for 5d9b5cb

Commits on Aug 1, 2020

• Add sm80 for CUDA >= 11 (NVIDIA#925)

  ptrblck

  authored
  5b53121

  Copy full SHA for 5b53121

Commits on Jun 1, 2020

• Add Pyprof removal warnings that point to new repo (NVIDIA#862)

  Show description for 097238f
  

  mcarilli

  and

  definitelynotmcarilli

  authored
  097238f

  Copy full SHA for 097238f

Commits on May 30, 2020

• Make separate apex option for distributed lamb

  thorjohnsen

  committed
  45388d4

  Copy full SHA for 45388d4

• Distributed LAMB optimizer

  thorjohnsen

  committed
  19892f1

  Copy full SHA for 19892f1

Commits on May 29, 2020

• Fuses dropout and softmax in backward pass, add bias support to CPP MHA, add additive mask support, separate Q/K/V parameters (NVIDIA#854)

  Show description for 6c2babf
  

  seryilmaz

  and

  Sukru Eryilmaz

  authored
  6c2babf

  Copy full SHA for 6c2babf

Commits on May 14, 2020

• Add FusedAdagrad (NVIDIA#822)

  ajtulloch

  authored
  3bae8c8

  Copy full SHA for 3bae8c8

Commits on Apr 23, 2020

• CUDAGenerator fix for #36026 (NVIDIA#801)

  Show description for 1f2aa91
  ptrblck

  authored
  1f2aa91

  Copy full SHA for 1f2aa91

Commits on Apr 22, 2020

• initial commit to add Multilayer Perceptron (MLP) extension (NVIDIA#790)

  FDecaYed

  authored
  71511fa

  Copy full SHA for 71511fa

Commits on Mar 23, 2020

• add l2norm source for FusedLAMB

  

  Kexin Yu

  committed
  a3ffb8a

  Copy full SHA for a3ffb8a

Commits on Mar 20, 2020

• extension name fix

  

  Kexin Yu

  committed
  b4c3201

  Copy full SHA for b4c3201

Pagination

Previous

Next