Commits

Breadcrumbs

History for

onmaster

User selector

All users

All time

Commit History

Commits on Feb 15, 2023

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

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  5146a68

  Copy full SHA for 5146a68

Commits on Feb 13, 2023

• [TTS/Torchhub] Expose HiFiGAN and FastPitch via TorchHub

  nv-kkudrynski

  committed
  afea561

  Copy full SHA for afea561

Commits on Feb 7, 2023

• [WaveGlow/PyT] Enable TorchScript

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  0e1c6c5

  Copy full SHA for 0e1c6c5

Commits on Jan 10, 2023

• [FastPitch/PyT] Add mixed English and Mandarin bilingual support

  

  MuyangDu

  authored and

  nv-kkudrynski

  committed
  c2bb3fe

  Copy full SHA for c2bb3fe

• [FastPitch/PyT] Bump container to 22.08, update perf results

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  72a15ee

  Copy full SHA for 72a15ee

Commits on Dec 13, 2022

• [HiFi-GAN/PyT] Whitelist data/cmudict in .dockerignore

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  56ce6c8

  Copy full SHA for 56ce6c8

Commits on Oct 27, 2022

• [HiFi-GAN/PyT] Import amp_C (apex) only when necessary

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  91c1de2

  Copy full SHA for 91c1de2

• [HiFi-GAN/PyT] Remove hyphen in path: HiFi-GAN -> HiFiGAN

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  ab3a0e4

  Copy full SHA for ab3a0e4

Commits on Oct 4, 2022

• [Speech/PyT] Save two dllogs: .json and .json.X

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  02e1ccb

  Copy full SHA for 02e1ccb

Commits on Oct 3, 2022

• [HiFi-GAN/PyT] Explicitly flush logs

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  66da9fd

  Copy full SHA for 66da9fd

Commits on Sep 14, 2022

• [FastPitch/PyT] Fix off-by-one in grad acc

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  a2f02eb

  Copy full SHA for a2f02eb

Commits on Sep 9, 2022

• [FastPitch/PyT] Resolve perf regression on DGX A100 + new perf tweaks

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  84b22b3

  Copy full SHA for 84b22b3

Commits on Sep 7, 2022

• [FastPitch/PyT][HiFi-GAN/PyT] Unidecoder: disable PyTorch warnings in all speech models

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  469df9b

  Copy full SHA for 469df9b

Commits on Sep 5, 2022

• [Tacotron2/PyT] Minor bug fixes

  

  grzegorz-k-karch

  authored and

  nv-kkudrynski

  committed
  47d783f

  Copy full SHA for 47d783f

• Merge: [Tacotron2/PyT] stabilize inference performance results for Tacotron 2

  nv-kkudrynski

  committed
  5e70f90

  Copy full SHA for 5e70f90

• [Tacotron2/PyT] stabilize inference performance results for Tacotron 2

  

  grzegorz-k-karch

  authored and

  nv-kkudrynski

  committed
  0422fa0

  Copy full SHA for 0422fa0

• [QuartzNet/PyT][Jasper/PyT][HiFi-GAN/PyT][FastPitch/PyT] Mute pyt is_namedtuple warning (1.13.0a0+d321be6)

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  206a46c

  Copy full SHA for 206a46c

Commits on Aug 16, 2022

• Merge: [FastPitch/PyT] Optimize CPU perf and remove GPU syncs

  nv-kkudrynski

  committed
  fc80de5

  Copy full SHA for fc80de5

• [FastPitch/PyT] Optimize CPU perf and remove GPU syncs

  

  timmoon10

  authored and

  nv-kkudrynski

  committed
  67469e9

  Copy full SHA for 67469e9

Commits on Aug 12, 2022

• [FastPitch/PyT] Add missing download_models.sh

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  69182fc

  Copy full SHA for 69182fc

Commits on Aug 9, 2022

• [HiFi-GAN/PyT][FastPitch/PyT] Fix wrong p-arpabet defaults, support NGC ckpts

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  d1b5b4d

  Copy full SHA for d1b5b4d

Commits on Jul 21, 2022

• [HiFi-GAN/PyT] Set missing CUDNN API V8 env vars in CI

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  8c7d66f

  Copy full SHA for 8c7d66f

Commits on Jul 7, 2022

• [FastPitch/PyT] Batched CTC loss

  

  timmoon10

  authored and

  nv-kkudrynski

  committed
  c91ce88

  Copy full SHA for c91ce88

Commits on Jul 5, 2022

• [HiFi-GAN/PyT] Remove debug code

  

  mikolajblaz

  authored and

  nv-kkudrynski

  committed
  d3ba35a

  Copy full SHA for d3ba35a

Commits on Jun 30, 2022

• [Tacotron2/PyT] Add train_loss to summary metrics

  

  grzegorz-k-karch

  authored and

  nv-kkudrynski

  committed
  96928fc

  Copy full SHA for 96928fc

Commits on Jun 27, 2022

• Merge: [CI/Speech] Add val losses to converge-partial

  nv-kkudrynski

  committed
  1f53a4d

  Copy full SHA for 1f53a4d

• [CI/Speech] Add val losses to converge-partial

  

  mikolajblaz

  authored and

  nv-kkudrynski

  committed
  25b5392

  Copy full SHA for 25b5392

• Add logging of units to dllogger metrics

  

  jan-golda

  authored and

  nv-kkudrynski

  committed
  e372b7b

  Copy full SHA for e372b7b

Commits on Jun 6, 2022

• Merge: [WaveGlow/PyT] fix for OOM bug

  nv-kkudrynski

  committed
  c16a623

  Copy full SHA for c16a623

• [WaveGlow/PyT] fix for OOM bug

  

  grzegorz-k-karch

  authored and

  nv-kkudrynski

  committed
  a52c44f

  Copy full SHA for a52c44f

• Merge: [FastPitch/PyT] Fix incorrect permute order in self-attention

  nv-kkudrynski

  committed
  58cdfff

  Copy full SHA for 58cdfff

• [FastPitch/PyT] Fix incorrect permute order in self-attention

  

  timmoon10

  authored and

  nv-kkudrynski

  committed
  86c9acf

  Copy full SHA for 86c9acf

Commits on May 31, 2022

• Merge: [FastPitch/PyT] Remove old checkpoints during training

  nv-kkudrynski

  committed
  148a9f2

  Copy full SHA for 148a9f2

• [FastPitch/PyT] Remove old checkpoints during training

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  8658e1b

  Copy full SHA for 8658e1b

• [Jasper/PyT][QuartzNet/PyT][FastPitch/PyT][HiFi-GAN/PyT] Monkey patch PyTorch state_dict() warnings

  

  alancucki

  authored and

  nv-kkudrynski

  committed
  0e279a3

  Copy full SHA for 0e279a3