Commits

Branch selector

 master

User selector

All users

All time

Commit History

Commits on Feb 25, 2022

• Add a --warmup_iters option to specify how many iterations are used as warmup

  desertfire

  committed
  a4b93bd

  Copy full SHA for a4b93bd

Commits on Feb 24, 2022

• Pass gradient_as_bucket_view=True to DDP to fix the NaN problem

  

  Jiewen Tan

  committed
  a36f4ed

  Copy full SHA for a36f4ed

• Add --fuser for specifying which fuser to use

  desertfire

  committed
  8be71cb

  Copy full SHA for 8be71cb

• Add --sync_every_iter to turn on synchronization at the end of each iteration

  desertfire

  committed
  2ebcb4c

  Copy full SHA for 2ebcb4c

Commits on Feb 23, 2022

• Move _ltc_init_ts_backend() to after "LTC_TS_CUDA" is set

  Show description for 37f9830
  desertfire

  committed
  37f9830

  Copy full SHA for 37f9830

• Add --ltc_summary for dumping LTC summary

  desertfire

  committed
  cda387e

  Copy full SHA for cda387e

Commits on Feb 18, 2022

• Move the CUDA sync and LTC sync at each iteration to the end of iterations

  Show description for 2bd3037
  desertfire

  committed
  2bd3037

  Copy full SHA for 2bd3037

Commits on Feb 17, 2022

• - Make DataLoader running in eager instead of lazy

  Show description for 2b7e27e
  desertfire

  committed
  2b7e27e

  Copy full SHA for 2b7e27e

Commits on Feb 16, 2022

• Enable DDP with LazyTensor

  

  Jiewen Tan

  committed
  b526e50

  Copy full SHA for b526e50

Commits on Feb 11, 2022

• Add a --ltc option to run Lazy Tensor

  Show description for e2ad817
  desertfire

  committed
  e2ad817

  Copy full SHA for e2ad817

Commits on Feb 2, 2022

• Merge: [Unet3d/TF] rename evaluation metrics

  nv-kkudrynski

  committed
  a43ffd0

  Copy full SHA for a43ffd0

• [Unet3d/TF] rename evaluation metrics

  

  mmarcinkiewicz

  authored and

  nv-kkudrynski

  committed
  f7f0761

  Copy full SHA for f7f0761

• [Convnets/TF] Added barrier at the end of multiprocess run to allow step synchronization

  

  hXl3s

  authored and

  nv-kkudrynski

  committed
  e1af187

  Copy full SHA for e1af187

Commits on Jan 31, 2022

• [nnUNet/PyT] New command line arguments; improved logging

  

  michal2409

  authored and

  nv-kkudrynski

  committed
  18e7756

  Copy full SHA for 18e7756

• [ELECTRA2/TF2] Provide optional output directory to save checkpoints

  

  sharathts

  authored and

  nv-kkudrynski

  committed
  11d1141

  Copy full SHA for 11d1141

Commits on Jan 28, 2022

• Merge: [Tacotron2/PyT][WaveGlow/PyT] Can set training random seed

  nv-kkudrynski

  committed
  cd91340

  Copy full SHA for cd91340

• [Tacotron2/PyT][WaveGlow/PyT] Can set training random seed

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  894d2c2

  Copy full SHA for 894d2c2

Commits on Jan 27, 2022

• [Internal/PyT/tacotron2] Convergence update

  

  grzegorz-k-karch

  authored and

  shakandrew

  committed
  c992690

  Copy full SHA for c992690

Commits on Jan 25, 2022

• Change DLLogger installation source

  

  shakandrew

  authored and

  nv-kkudrynski

  committed
  3a8068b

  Copy full SHA for 3a8068b

Commits on Jan 10, 2022

• Merge: [nnUNet/PyT] Fix DALI inference pipeline

  nv-kkudrynski

  committed
  589604d

  Copy full SHA for 589604d

• [nnUNet/PyT] Fix DALI inference pipeline

  

  michal2409

  authored and

  nv-kkudrynski

  committed
  45639af

  Copy full SHA for 45639af

• Merge: Update dllogger installation

  nv-kkudrynski

  committed
  0c3f7ab

  Copy full SHA for 0c3f7ab

• Merge: [DLRM/PyT] 21.10 container update with added graph support and BYOD capability

  nv-kkudrynski

  committed
  00100f2

  Copy full SHA for 00100f2

• [DLRM/PyT] 21.10 container update with added graph support and BYOD capability

  

  Tomasz Cheda

  authored and

  nv-kkudrynski

  committed
  4154bda

  Copy full SHA for 4154bda

Commits on Jan 5, 2022

• Update dllogger installation

  nv-dmajchrowski

  committed
  22a214e

  Copy full SHA for 22a214e

• [DLRM/TF2] Fix preprocessing and training Dockerfiles

  

  tgrel

  authored and

  nv-kkudrynski

  committed
  41cc33f

  Copy full SHA for 41cc33f

• [FastPitch/PyT] Update flags for inference

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  16c0b93

  Copy full SHA for 16c0b93

Commits on Dec 17, 2021

• [UNet/TF2] Add tf-trt and SavedModel tests. Remove profiling tests.

  

  mmarcinkiewicz

  authored and

  nv-kkudrynski

  committed
  ddbcd54

  Copy full SHA for ddbcd54

• [Tacotron2/PyT] Fix warnings, AMP state loading

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  7ce1754

  Copy full SHA for 7ce1754

• Merge: [QuartzNet/PyT][Jasper/PyT] Update requirements.txt

  nv-kkudrynski

  committed
  eda18ea

  Copy full SHA for eda18ea

• [QuartzNet/PyT][Jasper/PyT] Update requirements.txt

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  50aa2d6

  Copy full SHA for 50aa2d6

• Merge: [nnUNet/PyT] Remove profiling

  nv-kkudrynski

  committed
  fbef517

  Copy full SHA for fbef517

• [nnUNet/PyT] Remove profiling

  

  michal2409

  authored and

  nv-kkudrynski

  committed
  37dd0ff

  Copy full SHA for 37dd0ff

• Merge: [Efficientdet/PyT] Checkpoint URL fix

  nv-kkudrynski

  committed
  067d21e

  Copy full SHA for 067d21e

• [Efficientdet/PyT] Checkpoint URL fix

  

  subhankar-ghosh

  authored and

  nv-kkudrynski

  committed
  654ba04

  Copy full SHA for 654ba04

Pagination

Previous

Next