Commits

Breadcrumbs

History for

onmaster

User selector

All users

All time

Commit History

Commits on Dec 8, 2023

• [wav2vec2.0/PyT] Fix pip dependencies (librosa - numpy)

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  9dd9fcb

  Copy full SHA for 9dd9fcb

Commits on Dec 4, 2023

• Merge branch 'mwawrzos/nvbug/4393747' into 'internal/main'

  Show description for 0131db6
  nv-kkudrynski

  committed
  0131db6

  Copy full SHA for 0131db6

• [Jasper/PyT,QuartzNet/PyT] use the dafault DALI installation from the container

  

  mwawrzos

  authored and

  nv-kkudrynski

  committed
  34770bb

  Copy full SHA for 34770bb

Commits on Nov 13, 2023

• [DLRM/PyT] Fix np.bool API deprecation

  tgrel

  committed
  e52bcb0

  Copy full SHA for e52bcb0

Commits on Jun 30, 2023

• [Jasper/PyT, QuartzNet/PyT] Fix Ada L40 on 23.06 base container

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  8ed53a4

  Copy full SHA for 8ed53a4

Commits on May 29, 2023

• Merge: [NCF/PyT] Stop using deprecated apex AMP and apex DDP

  nv-kkudrynski

  committed
  2a7c251

  Copy full SHA for 2a7c251

• Merge: [DLRM/PyT] Stop using apex AMP and DDP

  nv-kkudrynski

  committed
  370a221

  Copy full SHA for 370a221

Commits on May 22, 2023

• [Jasper/PyT, QuartzNet/PyT] Update Pandas and Dali versions

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  9becdf8

  Copy full SHA for 9becdf8

Commits on May 15, 2023

• [Transformer/PyT] minor bugfix

  

  jbaczek

  authored and

  nv-kkudrynski

  committed
  7b89aed

  Copy full SHA for 7b89aed

Commits on Apr 24, 2023

• [NCF/PyT] Stop using deprecated apex AMP and apex DDP

  tgrel

  committed
  05ee986

  Copy full SHA for 05ee986

• [DLRM/PyT] Stop using apex AMP and DDP

  tgrel

  committed
  5bc69ca

  Copy full SHA for 5bc69ca

Commits on Apr 19, 2023

• [Jasper/PyT] Update torch.stft for PyTorch 2.0

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  29aaae3

  Copy full SHA for 29aaae3

Commits on Apr 11, 2023

• [TFT/PyT] Update Checkpoints

  

  IzzyPutterman

  authored and

  nv-kkudrynski

  committed
  a6c678e

  Copy full SHA for a6c678e

Commits on Mar 31, 2023

• [TFT/PyTorch] Move to nvFuser

  

  IzzyPutterman

  authored and

  nv-kkudrynski

  committed
  777d174

  Copy full SHA for 777d174

Commits on Mar 30, 2023

• [PyT/EffDet] Replace _six.string_classes with Actual Values

  

  suiyoubi

  authored and

  nv-kkudrynski

  committed
  0953478

  Copy full SHA for 0953478

Commits on Mar 24, 2023

• Merge: [BERT/PyT] set default bin size to 64 for phase2

  nv-kkudrynski

  committed
  441cf4b

  Copy full SHA for 441cf4b

• [BERT/PyT] set default bin size to 64 for phase2

  

  sharathts

  authored and

  nv-kkudrynski

  committed
  a0781f8

  Copy full SHA for a0781f8

• [ConvNets/PyT] Fix interpolation type from Image.* to InterpolationMode.*

  

  AdamRajfer

  authored and

  nv-kkudrynski

  committed
  a684cf0

  Copy full SHA for a684cf0

Commits on Mar 1, 2023

• [EffDet/PyT] Invoking CUDA synchronize() before Timing

  

  suiyoubi

  authored and

  nv-kkudrynski

  committed
  f613b7c

  Copy full SHA for f613b7c

Commits on Feb 23, 2023

• [Maskrcnn/PyT] Synchronize before reporting DLL time

  ShriyaRishab

  committed
  c44b7ad

  Copy full SHA for c44b7ad

Commits on Feb 15, 2023

• Merge: [GNMT/PyT] Added synchronization before collecting timers, switched to correct averaging when reporting avg throughput

  nv-kkudrynski

  committed
  ed28348

  Copy full SHA for ed28348

• Merge: [TXL/PyT] Added barriers when reporting time, switched to correct averaging when reporting avg throughput

  nv-kkudrynski

  committed
  7ddd062

  Copy full SHA for 7ddd062

• Merge: [Speech models/PyT] Update perf timers and cuda syncs

  nv-kkudrynski

  committed
  959c677

  Copy full SHA for 959c677

• [Speech models/PyT] Update perf timers and cuda syncs

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  5146a68

  Copy full SHA for 5146a68

• Merge: [BART/PyT] Add synchronize for benchmarking

  nv-kkudrynski

  committed
  c75cc20

  Copy full SHA for c75cc20

• Merge: [Transformer/PyTorch] Add torch.cuda.synchronize() calls

  nv-kkudrynski

  committed
  d46a356

  Copy full SHA for d46a356

• [Transformer/PyTorch] Add torch.cuda.synchronize() calls

  

  jbaczek

  authored and

  nv-kkudrynski

  committed
  efed242

  Copy full SHA for efed242

• Merge: [TFT/PyTorch] Added torch.cuda.synchronize() calls in perf meter

  nv-kkudrynski

  committed
  9868e23

  Copy full SHA for 9868e23

• [TFT/PyTorch] Added torch.cuda.synchronize() calls in perf meter

  

  jbaczek

  authored and

  nv-kkudrynski

  committed
  d190b25

  Copy full SHA for d190b25

• Merge: [DLRM/PyT] Add synchronization for more reliable benchmarking

  nv-kkudrynski

  committed
  3946f33

  Copy full SHA for 3946f33

• [DLRM/PyT] Add synchronization for more reliable benchmarking

  

  tgrel

  authored and

  nv-kkudrynski

  committed
  f5111db

  Copy full SHA for f5111db

• Merge: [nnUNet/PyT] Add torch.cuda.synchronize call while benchmarking

  nv-kkudrynski

  committed
  ff2f120

  Copy full SHA for ff2f120

Commits on Feb 13, 2023

• [BART/PyT] Add synchronize for benchmarking

  meatybobby

  committed
  a797214

  Copy full SHA for a797214

• [TXL/PyT] Added barriers when reporting time, switched to correct averaging when reporting avg throughput

  szmigacz

  committed
  8f82237

  Copy full SHA for 8f82237

• [GNMT/PyT] Added synchronization before collecting timers, switched to correct averaging when reporting avg throughput

  szmigacz

  committed
  327898a

  Copy full SHA for 327898a