Commits

History for

Commits on Jan 30, 2021

Pytorch Lightning Integration (SeanNaren#569 )

committedJan 30, 2021

Commits on Jul 14, 2020

Added hydra integration for inference/evaluation, added smoke tests, Added integration for transcribe.py, added pre-trained test with a refactor to use common utils. added fix for deprecated audio_…
sean.narenthiran
committedJul 14, 2020

Commits on Jul 4, 2020

Added ability to save checkpoints to GCS
sean.narenthiran
committedJul 4, 2020

Commits on Jul 3, 2020

Added config group for bidirectional/unidirectional model, fixed type checking on configs
sean.narenthiran
committedJul 3, 2020
Removed DictConfig inheritance and opted for no pretty print since this will be phased out before RC is released, added config yaml instructions
sean.narenthiran
committedJul 3, 2020

Commits on Jun 26, 2020

Added config group and default for optims, moved into config python script
sean.narenthiran
committedJun 26, 2020
Fixed import issue, added Enums for distributed backend/spectrogram window
sean.narenthiran
committedJun 26, 2020

Commits on Jun 25, 2020

Swapped to using Hydra dataclasses config groups
sean.narenthiran
committedJun 25, 2020

Commits on Jun 24, 2020

Moving towards a pypi package structure for adding integration tests
sean.narenthiran
committedJun 24, 2020

Commits on Jun 20, 2020

Integrated Hydra into the training script
sean.narenthiran
committedJun 20, 2020

Commits on Jun 4, 2020

Added AdamW optimizer
sean.narenthiran
committedJun 4, 2020

Commits on May 25, 2020

Fixed to benchmark.py, added suitable defaults reflected by release parameters

aadil-srivastava01
authored and
sean.narenthiran
committedMay 25, 2020

Commits on May 24, 2020

Refactor of code to fit torch elastic, improving save/loading of training state with state class, using torch elastic as multi-GPU/multi-Node training, additional functions to aid saving models, im…
sean.narenthiran
committedMay 24, 2020

Commits on May 22, 2020

Removed flag which was half implemented, added docs to Apex commands, small change to comment regarding flag
sean.narenthiran
committedMay 22, 2020

Commits on Apr 17, 2020

Modified labels into a list based on upstream ctcdecode changes
sean.narenthiran
committedApr 17, 2020

Commits on Feb 24, 2020

Added check to model to ensure we take off the distributed wrapper before saving
sean.narenthiran
committedFeb 24, 2020

Commits on Feb 20, 2020

Set loss-scale to 1 to prevent using dynamic scaling
sean.narenthiran
committedFeb 20, 2020

Commits on Feb 4, 2020

added spec_augment impl
pskrunner14
committedFeb 4, 2020

Commits on Jan 16, 2020

Fix gradient clipping amp compatibility, save and load amp state
marekrydlewski
committedJan 16, 2020

Commits on Aug 15, 2019

Merge branch 'master' into feature/V2
sean.narenthiran
committedAug 15, 2019

Commits on Aug 2, 2019

Use the same evaluation function in training/testing, ensure that we remove space characters before computing CER
sean.narenthiran
committedAug 2, 2019

Commits on Jul 25, 2019

Removed need for torchaudio, added improvements for APEX mixed precision, improved tuning for LMs, updated Docs
sean.narenthiran
committedJul 25, 2019

Commits on May 20, 2019

Fixed issue where continue_from option unconditionally overwrites best_wer.
jybaek
committedMay 20, 2019

Commits on May 10, 2019

Fixed Indent
jybaek
committedMay 10, 2019

Commits on Feb 7, 2019

fix continue path references
bnroths
committedFeb 7, 2019

Commits on Feb 1, 2019

Removed data time sync
Sean Naren
committedFeb 1, 2019

Commits on Jan 29, 2019

Major refactors to simplify code, support mixed precision, update dockerfile/README
sean.narenthiran
committedJan 29, 2019

Commits on Jan 7, 2019

Fixed error with transpose using newest torchaudio, added pytorch 1.0 changes
sean.narenthiran
committedJan 7, 2019

Commits on Jul 23, 2018

Fix "RuntimeError: Expected object of type torch.FloatTensor but found type torch.cuda.FloatTensor for argument SeanNaren#4 'other'" for File "train.py", line 270, in <module> optimizer.step()
shuieryin
committedJul 23, 2018

Commits on Jul 14, 2018

Added docs, changes to update server file, refactored options
sean.narenthiran
committedJul 14, 2018

Commits on Jul 12, 2018

Added ability to specify device IDs, added reduce operation for distribution to sync loss
sean.narenthiran
committedJul 12, 2018
Removed deprecated device-ids
sean.narenthiran
committedJul 12, 2018
Removed variables, added new masking layer, refactored training/loading of model
sean.narenthiran
committedJul 12, 2018

Commits on Jul 11, 2018

Changes to support variable length

SeanNaren
authored and
sean.narenthiran
committedJul 11, 2018

Commits on Jul 6, 2018

Update train.py
dalonlobo
committedJul 6, 2018