Commits

Breadcrumbs

History for

onmaster

User selector

All users

All time

Commit History

Commits on Aug 3, 2021

• Synpaflex french dataset tacotron2 and MB-MelGAN support

  samuel-lunii

  committed
  4a7d584

  Copy full SHA for 4a7d584

Commits on Nov 19, 2020

• 🦟 Update global batch-size for train-dataloder.

  dathudeptrai

  committed
  ac4508a

  Copy full SHA for ac4508a

• 🚲 Added gradient_accumulate_steps for all config, clearer note for batch_size parameter.

  dathudeptrai

  committed
  1ca581f

  Copy full SHA for 1ca581f

Commits on Nov 18, 2020

• 🐛 Output of melgan/mb-melgan should be float32 to prevent nan in mixed_precision.

  dathudeptrai

  committed
  518ee49

  Copy full SHA for 518ee49

Commits on Nov 7, 2020

• Fixed typo in MB-Melgan config

  OscarVanL

  committed
  940240a

  Copy full SHA for 940240a

Commits on Oct 21, 2020

• 🐛 fix training mb melgan in baker dataset

  

  yablon

  committed
  2474c79

  Copy full SHA for 2474c79

Commits on Oct 15, 2020

• 🏄 Fix MultiGPU utt_ids (#309).

  dathudeptrai

  committed
  5eee804

  Copy full SHA for 5eee804

Commits on Oct 13, 2020

• 📰 Add converted LibriTTS universal vocoder to pretrained model list; add Notes section

  ZDisket

  committed
  ede4c30

  Copy full SHA for ede4c30

Commits on Sep 26, 2020

• 😶 Do not use mixed precision when Melgan/mb-melgan discriminator enable.

  dathudeptrai

  committed
  c4e723a

  Copy full SHA for c4e723a

Commits on Sep 7, 2020

• 🤰 Save utt_id rather than idx.

  dathudeptrai

  committed
  a454ea3

  Copy full SHA for a454ea3

Commits on Aug 27, 2020

• 🐟 Change logging of pretrained load; add sapace

  ZDisket

  committed
  b900554

  Copy full SHA for b900554

• 📰 Document --pretrained argument in READMEs

  ZDisket

  committed
  8f247fb

  Copy full SHA for 8f247fb

• Add --pretrained argument to all training scripts; normalize spaces

  ZDisket

  committed
  3566e5e

  Copy full SHA for 3566e5e

Commits on Aug 14, 2020

• 🇨🇳 Ready to merge.

  dathudeptrai

  committed
  0a9d774

  Copy full SHA for 0a9d774

Commits on Aug 11, 2020

• Merge branch 'master' into dev/chinese_example

  dathudeptrai

  authored
  b688773

  Copy full SHA for b688773

• update chinese example

  azraelkuan

  committed
  27004f4

  Copy full SHA for 27004f4

Commits on Aug 9, 2020

• 😒 Fix typo.

  dathudeptrai

  committed
  ca9f678

  Copy full SHA for ca9f678

Commits on Aug 5, 2020

• Merge branch 'master' into dev/chinese_example

  dathudeptrai

  authored
  b5bc31c

  Copy full SHA for b5bc31c

• fix parameters name

  azraelkuan

  committed
  0279ca6

  Copy full SHA for 0279ca6

• add model_type

  azraelkuan

  committed
  d44892e

  Copy full SHA for d44892e

• 💉 Update README, pretrained link for korean.

  dathudeptrai

  committed
  e94e780

  Copy full SHA for e94e780

Commits on Aug 3, 2020

• merge changes

  azraelkuan

  committed
  0ab26ac

  Copy full SHA for 0ab26ac

• Merge branch 'master' into kss

  dathudeptrai

  authored
  722c171

  Copy full SHA for 722c171

Commits on Aug 2, 2020

• 🔧 Bug 🐛 wrong name in melgan training script.

  dathudeptrai

  committed
  2271faf

  Copy full SHA for 2271faf

• Merge branch 'master' into kss

  Show description for 96e5bc9
  crux153

  committed
  96e5bc9

  Copy full SHA for 96e5bc9

Commits on Jul 30, 2020

• 🚀 Add TFAutoModel and AutoConfig.

  dathudeptrai

  committed
  b4497d6

  Copy full SHA for b4497d6

Commits on Jul 22, 2020

• merge changes

  azraelkuan

  committed
  8472437

  Copy full SHA for 8472437

• Merge branch 'master' into kss

  Show description for 5b546d9
  crux153

  committed
  5b546d9

  Copy full SHA for 5b546d9

Commits on Jul 21, 2020

• Fix decode_mb_melgan.py

  crux153

  committed
  70c1658

  Copy full SHA for 70c1658

• ✍ Move all mixed_precision optimization to base_trainer.

  dathudeptrai

  committed
  a280b61

  Copy full SHA for a280b61

• ✍ Update README for clarify how to resume training progress.

  dathudeptrai

  committed
  28da381

  Copy full SHA for 28da381

Commits on Jul 20, 2020

• ✍ Update ReadMe for comput_per_example_loss on each model.

  dathudeptrai

  committed
  d59e652

  Copy full SHA for d59e652

Commits on Jul 17, 2020

• ✍ Update README for MultiGPU traning, add doc for Compute per example loss function.

  dathudeptrai

  committed
  a7778c5

  Copy full SHA for a7778c5

• 🚀 Change default init lr for melgan.stft/mb_melgan to 0.0005, it improve the loss significantly.

  dathudeptrai

  committed
  b10a6e1

  Copy full SHA for b10a6e1

• 🚀 Support MultiGPU/SiggleGPU in one base_trainer, Refactor project.

  dathudeptrai

  committed
  5c32494

  Copy full SHA for 5c32494