Tutorial Config Bug Fix #2232
Merged
Tutorial Config Bug Fix #2232
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Signed-off-by: fayejf <fayejf07@gmail.com>
karpnv
pushed a commit
to karpnv/NeMo
that referenced
this pull request
May 21, 2021
Signed-off-by: fayejf <fayejf07@gmail.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Карпов Николай Вячеславович <karpov.n.vyac@sberbank.ru>
michalivne
pushed a commit
to michalivne/NeMo
that referenced
this pull request
Jun 23, 2021
Signed-off-by: fayejf <fayejf07@gmail.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com>
mousebaiker
pushed a commit
to mousebaiker/NeMo
that referenced
this pull request
Jul 8, 2021
Signed-off-by: fayejf <fayejf07@gmail.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com>
MaximumEntropy
added a commit
that referenced
this pull request
Jul 14, 2021
* Itn add classes (#2141) * move do_training flag to config Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * added telephone to itn Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * add telephone and email to itn Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * ASR + NLP Doc Fixes (#2136) * Preserve the tokenizer config for ASR Signed-off-by: smajumdar <titu1994@gmail.com> * Correct nlp docs Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Removing graphsurgeon optional dependency, improving import error rep… (#2144) * Removing graphsurgeon optional dependency, improving import error reporting Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixing scope error Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix FilterbankFeatures eval nondeterminism. (#2146) Signed-off-by: PiotrDabkowski <piodrus@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix the docs. (#2148) Signed-off-by: Micha Livne <mlivne@nvidia.com> * Text processing refactor (#2149) * removed graphutils, suppletive, data_loader_utils from itn to be reused from tn Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * inheriting itn from tn, thus removing redundancy Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * cleaned whitelist Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * lgtm fix Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update how artifacts work (#2138) * Update how artifacts work Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * fixing some tests Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * fix more tests Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * add __init__ to tests to make them discoverable Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * empty src support Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * updates plust unittest Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * add copyright check Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * copyright header Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * fix style Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * handle hashed megatron checkpoint version in nlp restore_from Signed-off-by: ericharper <complex451@gmail.com> * add _MODEL_RESTORE_PATH to AppState Signed-off-by: ericharper <complex451@gmail.com> * get rid of global folder caching Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * double register - warning instead of exception Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * Add asr spe tests Signed-off-by: smajumdar <titu1994@gmail.com> * Pop out asr wpe pre-registered value Signed-off-by: smajumdar <titu1994@gmail.com> * Correct ASR tests and paths Signed-off-by: smajumdar <titu1994@gmail.com> * Correct tokenizer saving Signed-off-by: smajumdar <titu1994@gmail.com> * Correct ASR tests Signed-off-by: smajumdar <titu1994@gmail.com> * Correct ASR bpe mixin Signed-off-by: smajumdar <titu1994@gmail.com> * Patch up backward compatibility Signed-off-by: smajumdar <titu1994@gmail.com> * update register_bert_model Signed-off-by: ericharper <complex451@gmail.com> * update all get_lm_model calls Signed-off-by: ericharper <complex451@gmail.com> * return None if src not found Signed-off-by: ericharper <complex451@gmail.com> * handle case with no tokenizer Signed-off-by: ericharper <complex451@gmail.com> * do not add another hash is using tarfile_artifacts Signed-off-by: ericharper <complex451@gmail.com> * add return_none flag, update doc string Signed-off-by: ericharper <complex451@gmail.com> * update default behavior of register_artifact for NLPModel Signed-off-by: ericharper <complex451@gmail.com> * change kwarg name to verify_src_exists Signed-off-by: ericharper <complex451@gmail.com> * use cfg instead of _cfg Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * some cleanups Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Co-authored-by: ericharper <complex451@gmail.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Language model refactoring (#2120) * fixed branch in IR tutorial Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * bucketing tarred dataset for lm training Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * updated global rank Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * perplexity update Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * refactor lm to be campatible with latest nmt Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * perplexity change Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * removed obsolete config Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * added sequence perplexity Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * added non-smoothed CE loss for validation Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * unified sentence dataset, torchmetrics for sequence perplexity Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * translate_ddp refactor Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Signed-off-by: Micha Livne <mlivne@nvidia.com> * [NMT] Multi-validation Patch (#2150) * rename dl index 0 loss and sacrebleu for backwards compatibility Signed-off-by: ericharper <complex451@gmail.com> * eval -> val/tst Signed-off-by: ericharper <complex451@gmail.com> * instantiate torchmetrics after instantiating dataloaders Signed-off-by: ericharper <complex451@gmail.com> * bug Signed-off-by: ericharper <complex451@gmail.com> * remove debugging log Signed-off-by: ericharper <complex451@gmail.com> * remove debugging log Signed-off-by: ericharper <complex451@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * bumping version to 1.0.0 Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fixed the num_samples of text classification model. (#2152) Signed-off-by: Vahid <vnoroozi@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix for electronic (#2153) * fix for electronic Signed-off-by: ekmb <ebakhturina@nvidia.com> * special symbols added Signed-off-by: ekmb <ebakhturina@nvidia.com> * restrict symbols list Signed-off-by: ekmb <ebakhturina@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * FastSpeech 2 Test & Docs (#2143) * Add FS2 data loading test Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * TTS docs update for FastSpeech 2 Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Style fix for FS2 dataset test Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Fix transpose typo Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Minor patch for translate_ddp (#2155) * Patch for backtranslation in lm dataset Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * One more fix Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Entity linking (#2050) * Started adding SAP dataset Signed-off-by: Virginia Adams <vadams@nvidia.com> * Delete .lm_bert_dataset.py.swp Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added dataset and loss Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added entity linking encoder model Signed-off-by: Virginia Adams <vadams@nvidia.com> * Can build and use index from pubmedbert model Signed-off-by: Virginia Adams <vadams@nvidia.com> * checked boolean logic in build_index.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * End to end tested all functionality Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed val loss none at end of validation Signed-off-by: Virginia Adams <vadams@nvidia.com> * Started adding demo entity linking notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * adding in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * added call to entitylinking classes in __init__.py files Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added eval code to notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * Adding unfinished notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * Cleaned up example dir Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed recap commands Signed-off-by: Virginia Adams <vadams@nvidia.com> * added model typing and tiny data tar Signed-off-by: Virginia Adams <vadams@nvidia.com> * Adding tiny data zip Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated tiny example config data path Signed-off-by: Virginia Adams <vadams@nvidia.com> * Notebook demo works Signed-off-by: Virginia Adams <vadams@nvidia.com> * Changed training epochs Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed output from training and install cells Signed-off-by: Virginia Adams <vadams@nvidia.com> * changed code formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Started doc string for new functions Signed-off-by: Virginia Adams <vadams@nvidia.com> * Updated data_preprocessing to save to data_dir Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed comment in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * Update data_preprocessing.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated nemo typing imports Signed-off-by: Virginia Adams <vadams@nvidia.com> * about to rebase Signed-off-by: Virginia Adams <vadams@nvidia.com> * added back umls_dataset_processing.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed example data Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed typos in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed lgtm-com issues Signed-off-by: Virginia Adams <vadams@nvidia.com> * added copyright headers Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed import and copyright headers Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed formatting changes 2 Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed test formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added __init__.py for model and dataset Signed-off-by: Virginia Adams <vadams@nvidia.com> * loading newline file returns data_dir now Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed conf notebook and deleted comment Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * Updated Jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed file path Signed-off-by: Virginia Adams <vadams@nvidia.com> * Changed Jenkins pipeline order Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed Jenkins datapath... again... Signed-off-by: Virginia Adams <vadams@nvidia.com> * Made most review changes Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed copy right Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated unit test to wget config Signed-off-by: Virginia Adams <vadams@nvidia.com> * reverted test file back Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added project dir to jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * defined config in unit test Signed-off-by: Virginia Adams <vadams@nvidia.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Correct branch version for v1.0.0 (#2157) * Correct branch version Signed-off-by: smajumdar <titu1994@gmail.com> * Correct Jenkinsfile Signed-off-by: smajumdar <titu1994@gmail.com> * Update rst files Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * switch CI back to main Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fixed the docs. (#2156) Signed-off-by: Micha Livne <mlivne@nvidia.com> * Make Hifigan jittable (#2159) * FastSpeech 2 Test & Docs (#2143) * Add FS2 data loading test Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * TTS docs update for FastSpeech 2 Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Style fix for FS2 dataset test Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Fix transpose typo Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Jason <jasoli@nvidia.com> * Entity linking (#2050) * Started adding SAP dataset Signed-off-by: Virginia Adams <vadams@nvidia.com> * Delete .lm_bert_dataset.py.swp Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added dataset and loss Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added entity linking encoder model Signed-off-by: Virginia Adams <vadams@nvidia.com> * Can build and use index from pubmedbert model Signed-off-by: Virginia Adams <vadams@nvidia.com> * checked boolean logic in build_index.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * End to end tested all functionality Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed val loss none at end of validation Signed-off-by: Virginia Adams <vadams@nvidia.com> * Started adding demo entity linking notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * adding in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * added call to entitylinking classes in __init__.py files Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added eval code to notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * Adding unfinished notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * Cleaned up example dir Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed recap commands Signed-off-by: Virginia Adams <vadams@nvidia.com> * added model typing and tiny data tar Signed-off-by: Virginia Adams <vadams@nvidia.com> * Adding tiny data zip Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated tiny example config data path Signed-off-by: Virginia Adams <vadams@nvidia.com> * Notebook demo works Signed-off-by: Virginia Adams <vadams@nvidia.com> * Changed training epochs Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed output from training and install cells Signed-off-by: Virginia Adams <vadams@nvidia.com> * changed code formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Started doc string for new functions Signed-off-by: Virginia Adams <vadams@nvidia.com> * Updated data_preprocessing to save to data_dir Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed comment in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * Update data_preprocessing.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated nemo typing imports Signed-off-by: Virginia Adams <vadams@nvidia.com> * about to rebase Signed-off-by: Virginia Adams <vadams@nvidia.com> * added back umls_dataset_processing.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed example data Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed typos in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed lgtm-com issues Signed-off-by: Virginia Adams <vadams@nvidia.com> * added copyright headers Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed import and copyright headers Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed formatting changes 2 Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed test formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added __init__.py for model and dataset Signed-off-by: Virginia Adams <vadams@nvidia.com> * loading newline file returns data_dir now Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed conf notebook and deleted comment Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * Updated Jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed file path Signed-off-by: Virginia Adams <vadams@nvidia.com> * Changed Jenkins pipeline order Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed Jenkins datapath... again... Signed-off-by: Virginia Adams <vadams@nvidia.com> * Made most review changes Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed copy right Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated unit test to wget config Signed-off-by: Virginia Adams <vadams@nvidia.com> * reverted test file back Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added project dir to jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * defined config in unit test Signed-off-by: Virginia Adams <vadams@nvidia.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> * switch CI back to main Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * Make Hifigan jittable Signed-off-by: Ryan Leary <rleary@nvidia.com> * Remove vestigial debugging printout Signed-off-by: Ryan Leary <rleary@nvidia.com> * Add export forward and fix style Signed-off-by: Ryan Leary <rleary@nvidia.com> * Fix load_state_dict override for arbitrary layers Signed-off-by: Ryan Leary <rleary@nvidia.com> Co-authored-by: Jocelyn <jocelynh@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Jason <jasoli@nvidia.com> Co-authored-by: vadam5 <78445382+vadam5@users.noreply.github.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Co-authored-by: Ryan Leary <rleary@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix version (#2162) Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Megatron nb size reduced (#2163) * notebook size reduced Signed-off-by: ekmb <ebakhturina@nvidia.com> * notebook size reduced Signed-off-by: ekmb <ebakhturina@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update spectral clustering method (#2158) * update spectral clustering method Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * update Jenkins File Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * threshold fix by reducing window length for shorter embs Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * grammar fixes Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * CR update Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * paper reference Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * improve docstring for yaml Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * Doc fixes Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * revert (#2167) Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Limit Pytorch lightning release (#2170) Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * token classification models artifacts update (#2169) * artifacts update Signed-off-by: ekmb <ebakhturina@nvidia.com> * artifacts update Signed-off-by: ekmb <ebakhturina@nvidia.com> * jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> * fix for model restoration Signed-off-by: ekmb <ebakhturina@nvidia.com> * typos fix + jenkins dir update Signed-off-by: ekmb <ebakhturina@nvidia.com> * typo Signed-off-by: ericharper <complex451@gmail.com> * update jenkins branch Signed-off-by: ericharper <complex451@gmail.com> * add && Signed-off-by: ericharper <complex451@gmail.com> * jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> * jenkins disable Signed-off-by: ekmb <ebakhturina@nvidia.com> * revert jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> * jenkins disable Signed-off-by: ekmb <ebakhturina@nvidia.com> * revert jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> Co-authored-by: ericharper <complex451@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix to always_save_nemo (#2174) * Initial attempt at always_save_nemo fix Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * updated path before saving in exp manager, fixed bug when handling tarfile artifacts Signed-off-by: ericharper <complex451@gmail.com> * Add test with always_save_nemo to exp_manager Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * Style fixes Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * update jenkins branch Signed-off-by: ericharper <complex451@gmail.com> * check for nemo: Signed-off-by: ericharper <complex451@gmail.com> * check for nemo: Signed-off-by: ericharper <complex451@gmail.com> * check for nemo: Signed-off-by: ericharper <complex451@gmail.com> Co-authored-by: ericharper <complex451@gmail.com> Co-authored-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix typo (#2179) Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Make itn tests optional (#2173) * Limit Pytorch lightning release Signed-off-by: smajumdar <titu1994@gmail.com> * Add final two checks Signed-off-by: smajumdar <titu1994@gmail.com> Co-authored-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * First Revision of TTS Docs and Notebooks Update for 1.0 (#2166) * squash Signed-off-by: Jason <jasoli@nvidia.com> * notebook fixes Signed-off-by: Jason <jasoli@nvidia.com> * notebook fixes Signed-off-by: Jason <jasoli@nvidia.com> * typos Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * add more alternatives of 0 for telephone (#2171) Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Acc tn (#2180) * make tn cardinal faster Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * add number far Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * add test Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix lgtm Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix lgtm Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * [DOCS] NLP Model parallel, NMT multi-val, CORE register artifacts (#2168) * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Change label smoothing prob to reduce chance of test failure (#2184) Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Eric Harper <complex451@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add FS2 checkpoint links to docs and inference notebook (#2181) * Add FS2 checkpoint links to docs and inference notebook Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Remove empty cell from TTS notebook Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update ptl to 1.3 on main branch (#2178) * Update PTL Signed-off-by: smajumdar <titu1994@gmail.com> * Begin update to Pytorch Lightning 1.3.x Signed-off-by: smajumdar <titu1994@gmail.com> * Formatting Signed-off-by: smajumdar <titu1994@gmail.com> * style Signed-off-by: ericharper <complex451@gmail.com> * Formatting Signed-off-by: smajumdar <titu1994@gmail.com> * minor fix Signed-off-by: Jason <jasoli@nvidia.com> * minor fix Signed-off-by: Jason <jasoli@nvidia.com> * get testing attribute from trainer Signed-off-by: ericharper <complex451@gmail.com> * update init_ddp_connection override Signed-off-by: ericharper <complex451@gmail.com> * update attribute Signed-off-by: ericharper <complex451@gmail.com> * add barrier after load checkpoint in megatron Signed-off-by: ericharper <complex451@gmail.com> * remove barrier Signed-off-by: ericharper <complex451@gmail.com> * update last naming Signed-off-by: Jason <jasoli@nvidia.com> Co-authored-by: ericharper <complex451@gmail.com> Co-authored-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * SDE updates (#2187) * Added updates to SDE: - support for external vocabulary (to detect OOV words) - support for offset field (for segmented long recordings) - UI improvements Signed-off-by: Vitaly Lavrukhin <vlavrukhin@nvidia.com> * Refactored diff in SDE Signed-off-by: Vitaly Lavrukhin <vlavrukhin@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add TTS aligner and improved version of g2p for vocabs.Phonemes, small improvement in TalkNet (#2189) * add first version of aligner Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> * aligner docs, new g2p version, fix bugs in talknet Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> * update docs and remove lj related code Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> * fix style Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> * fix import Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * set the default of nodessplitter to None. (#2190) Signed-off-by: Vahid <vnoroozi@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * NMT fixes (#2194) * minor fixes Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * minor bugfixes Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Store mappings file in .nemo for FS2 model (#2196) * Store mappings file in .nemo for FS2 model Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Add error enforcing mappings file during training (FS2) Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add support to change the SE context window of ConvASREncoder (#2193) * Add support for changing context window on the fly Signed-off-by: smajumdar <titu1994@gmail.com> * Add support to change the SE context window of ConvASREncoder Signed-off-by: smajumdar <titu1994@gmail.com> * Add ability to skip config updating Signed-off-by: smajumdar <titu1994@gmail.com> * Switch to mixin based API Signed-off-by: smajumdar <titu1994@gmail.com> * Update docs and api for ASRModuleMixin Signed-off-by: smajumdar <titu1994@gmail.com> * Change print to logging.info Signed-off-by: smajumdar <titu1994@gmail.com> * Correct stride level when computing context window Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add a CI test for doing inference with an NMT model trained with Pre-LN (#2198) * Change label smoothing prob to reduce chance of test failure Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * Add Pre-LN inference test to Jenkinsfile Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * Separate tests for training and NMT inference Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix ipywidgets error in asr notebook (#2199) Added `ipywidgets` to avoid `ImportError: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html` error. Signed-off-by: Derek Chia <derek@derekchia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * metrics fix (#2202) * metrics fix Signed-off-by: ekmb <ebakhturina@nvidia.com> * metrics reset for punct model Signed-off-by: ekmb <ebakhturina@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * readme and minor improvements (#2203) * readme and minor improvements Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * vad threshold update Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix text processing docs (#2195) * fix text processing docs Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix name Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * add guard to pynini import Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix bug in SpecCutout (#2201) Signed-off-by: Robert Bracco <robertbracco1@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix bug in SpecCutout (#2201) (#2205) Signed-off-by: Robert Bracco <robertbracco1@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: smajumdar <titu1994@gmail.com> Co-authored-by: Robert Bracco <47190785+rbracco@users.noreply.github.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Set seed before generating random tensors in NMT test (#2206) * Change label smoothing prob to reduce chance of test failure Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * Set seed before generating tensors Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> Signed-off-by: Micha Livne <mlivne@nvidia.com> * ASR patches for v1.0.0 (#2207) * Multiple updates to RNNT add initialization Signed-off-by: smajumdar <titu1994@gmail.com> * Correct name of initilization Signed-off-by: smajumdar <titu1994@gmail.com> * Update dockerignore Signed-off-by: smajumdar <titu1994@gmail.com> * Fix RNNT WER calculation Signed-off-by: smajumdar <titu1994@gmail.com> * Address comments Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Multilingual training for NMT (#2160) * mnmt on fresh main Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * push for test Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * debug Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * check Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * cleanup Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * minor fix Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * more minor fixes Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * fix for test Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * fix list size error Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * multilingual in infer Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * changes Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * tar creation with multilingual Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * fix Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * changes + parallelism + bug fix Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * small fix Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * multilingual preprocessor fix Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * globally unique fragment names in tarred dataset Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * minor changes Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * rm load_from_cached_dataset Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * minor config change Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * rm unsued import Signed-off-by: Abhinav Khattar <aklife97@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Remove memory leak from ASR notebook + update model notebook (#2213) * ASR patches for v1.0.0 (#2207) * Multiple updates to RNNT add initialization Signed-off-by: smajumdar <titu1994@gmail.com> * Correct name of initilization Signed-off-by: smajumdar <titu1994@gmail.com> * Update dockerignore Signed-off-by: smajumdar <titu1994@gmail.com> * Fix RNNT WER calculation Signed-off-by: smajumdar <titu1994@gmail.com> * Address comments Signed-off-by: smajumdar <titu1994@gmail.com> * Correct model notebook to log the loss and correctly assign keys Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * replace names in vad tutorials (#2220) Signed-off-by: fayejf <fayejf07@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix the versioning name. (#2209) * fix the versioning name. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Made version None. Signed-off-by: Vahid <vnoroozi@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Enabled passing kwargs to export() (#2175) * Enabled passing kwargs to export() Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixing style; changed Classifier input_example to new extended syntax Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixed order of forward() call in export Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixing style Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update g2p: ambigious ignore, flag for skipping seq2seq (#2223) Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update TTS notebook with TalkNet inference (#2133) * Update TTS notebook with TalkNet inference. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Update TTS Notebook. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Update TTS TN Training Notebook. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Fix TN paper link. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Remove branch updaing TODOs. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update speaker notebooks (#2224) Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Support symlinked files (#2216) Signed-off-by: Anas Abou Allaban <aabouallaban@pm.me> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Set strict=True everywhere by default. (#2225) Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * set strict=True in nlp_model (#2227) Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * set strict=False for model parallel examples Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Make Text processing installation optional via reinstall.sh (#2226) * Make Text processing installation optional via reinstall.sh Signed-off-by: smajumdar <titu1994@gmail.com> * Support both success and failure states Signed-off-by: smajumdar <titu1994@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Transformer final norm preln (#2197) * fix pre_ln final norm Signed-off-by: fayejf <fayejf07@gmail.com> * style fix Signed-off-by: fayejf <fayejf07@gmail.com> * bug fixed Signed-off-by: fayejf <fayejf07@gmail.com> * bugfix post_ln Signed-off-by: fayejf <fayejf07@gmail.com> * update and add pre_ln_final_norm Signed-off-by: fayejf <fayejf07@gmail.com> * style fix Signed-off-by: fayejf <fayejf07@gmail.com> * fix for unit test Signed-off-by: fayejf <fayejf07@gmail.com> * rename final_norm to final_layer_norm Signed-off-by: fayejf <fayejf07@gmail.com> * bug fix Signed-off-by: fayejf <fayejf07@gmail.com> * tiny fix Signed-off-by: fayejf <fayejf07@gmail.com> * fix and improve Signed-off-by: fayejf <fayejf07@gmail.com> * tiny fix Signed-off-by: fayejf <fayejf07@gmail.com> * Patch for NMT to allow loading old modlels trained with pre-LN Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update models and notebook for 1.0 (#2211) * update models Signed-off-by: Jason <jasoli@nvidia.com> * updates Signed-off-by: Jason <jasoli@nvidia.com> * fix Signed-off-by: Jason <jasoli@nvidia.com> * add links Signed-off-by: Jason <jasoli@nvidia.com> * fixes Signed-off-by: Jason <jasoli@nvidia.com> * style Signed-off-by: Jason <jasoli@nvidia.com> * update checkpoints Signed-off-by: Jason <jasoli@nvidia.com> * rename Signed-off-by: Jason <jasoli@nvidia.com> * fixes Signed-off-by: Jason <jasoli@nvidia.com> * lgtm Signed-off-by: Jason <jasoli@nvidia.com> * fix loading waveglow Signed-off-by: Jason <jasoli@nvidia.com> * typo Signed-off-by: Jason <jasoli@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update_metrics_classification_models (#2228) Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Co-authored-by: fayejf <36722593+fayejf@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Data loader for seq of label model (#2084) * feature to seq label data loader Signed-off-by: fayejf <fayejf07@gmail.com> * style fix Signed-off-by: fayejf <fayejf07@gmail.com> * small fix Signed-off-by: fayejf <fayejf07@gmail.com> * update tl to be length of seq label Signed-off-by: fayejf <fayejf07@gmail.com> * style fix Signed-off-by: fayejf <fayejf07@gmail.com> * tiny bug fix Signed-off-by: fayejf <fayejf07@gmail.com> * small updates Signed-off-by: fayejf <fayejf07@gmail.com> * updates for review feedback Signed-off-by: fayejf <fayejf07@gmail.com> * style fix Signed-off-by: fayejf <fayejf07@gmail.com> * explain seq_label Signed-off-by: fayejf <fayejf07@gmail.com> * fix lgtm Signed-off-by: fayejf <fayejf07@gmail.com> * small updates Signed-off-by: fayejf <fayejf07@gmail.com> * improve as discussed Signed-off-by: fayejf <fayejf07@gmail.com> * add docstring Signed-off-by: fayejf <fayejf07@gmail.com> Co-authored-by: Nithin Rao <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix comments (#2236) * fix comments Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix typo Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * add paper ref to sgdqa model doc (#2233) * add paper ref to sgdqa model doc Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix comments Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Move ConcatDataset to common (#2237) * move concatdataset to common Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * var name change Signed-off-by: Abhinav Khattar <aklife97@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * audio based normalization (#2231) * squash norm_audio Signed-off-by: ekmb <ebakhturina@nvidia.com> * add missing files Signed-off-by: ekmb <ebakhturina@nvidia.com> * style Signed-off-by: ekmb <ebakhturina@nvidia.com> * unit tests added, docstrings fixed Signed-off-by: ekmb <ebakhturina@nvidia.com> * fix lgtm errors Signed-off-by: ekmb <ebakhturina@nvidia.com> * debug jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> * debug jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> * signature update Signed-off-by: ekmb <ebakhturina@nvidia.com> * set deterministic default Signed-off-by: ekmb <ebakhturina@nvidia.com> * add more test cases Signed-off-by: ekmb <ebakhturina@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * bug fix config (#2232) Signed-off-by: fayejf <fayejf07@gmail.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Alias Swish to SiLU (#2239) * Alias Swish to SiLU and move activations to inplace execution if possible Signed-off-by: smajumdar <titu1994@gmail.com> * Remove unused import Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update README.rst Signed-off-by: Micha Livne <mlivne@nvidia.com> * Offline asr notebook bug fix (#2242) * fix Signed-off-by: fayejf <fayejf07@gmail.com> * install Signed-off-by: fayejf <fayejf07@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix docstring (#2244) * fix comments Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix doc string Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update "last" Checkpoint (#2241) * fix Signed-off-by: Jason <jasoli@nvidia.com> * change Signed-off-by: Jason <jasoli@nvidia.com> * fix Signed-off-by: Jason <jasoli@nvidia.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add pretrained model stt_es_citrinet_512 (#2247) Signed-off-by: Elena Rastorgueva <erastorgueva@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * [BUGFIX] Only process tarfile artifacts when model was restored from tarfile (#2250) * process tarfile artifacts only if model is being restored Signed-off-by: ericharper <complex451@gmail.com> * process tarfile artifacts only if model was restored from a tarfile Signed-off-by: ericharper <complex451@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Log average metrics for Multi-validation in NMT (#2251) * add avg metrics NMT Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * name change Signed-off-by: Abhinav Khattar <aklife97@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update Primer notebook (#2258) Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fixed Bug 3310780 and 3310799 (#2264) Signed-off-by: Virginia Adams <vadams@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Support multiple models being instantiated in same execution scope (#2245) * Support multiple models being instantiated in same execution scope Signed-off-by: smajumdar <titu1994@gmail.com> * Fix tests Signed-off-by: smajumdar <titu1994@gmail.com> * Add locks to methods in appstate Signed-off-by: smajumdar <titu1994@gmail.com> * Perform locks only on write operations Signed-off-by: smajumdar <titu1994@gmail.com> * Correct deadlock issue Signed-off-by: smajumdar <titu1994@gmail.com> * Add more tests Signed-off-by: smajumdar <titu1994@gmail.com> * Add test for multi save and remove patch to change save type Signed-off-by: smajumdar <titu1994@gmail.com> * Update app state to preserve gidx of previous token Signed-off-by: smajumdar <titu1994@gmail.com> * Correct restoration logic for tarfiles Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * ASR Refactoring (#2240) * Refactor out the preprocessing from ASR into common Signed-off-by: smajumdar <titu1994@gmail.com> * Correct nltk issue with vocabs.py for clusters Signed-off-by: smajumdar <titu1994@gmail.com> * Add typing information to SpecAugment and SpecCutout Signed-off-by: smajumdar <titu1994@gmail.com> * Reorganize parts directory Signed-off-by: smajumdar <titu1994@gmail.com> * Refactor parts submodules, add __init__ to few important parts Signed-off-by: smajumdar <titu1994@gmail.com> * Update docs for new path to parts Signed-off-by: smajumdar <titu1994@gmail.com> * Cherry pick PR https://github.com/NVIDIA/NeMo/pull/2219 Signed-off-by: smajumdar <titu1994@gmail.com> * Add header for preprocessing commons Signed-off-by: smajumdar <titu1994@gmail.com> * Fix style of tests Signed-off-by: smajumdar <titu1994@gmail.com> * Add forced update of configs for train-val-test ds to new labels tests Signed-off-by: smajumdar <titu1994@gmail.com> * Update path to FilterbankFeatures for TTS Signed-off-by: smajumdar <titu1994@gmail.com> * Add an alias file for backward compatibility Signed-off-by: smajumdar <titu1994@gmail.com> * Add an alias file for backward compatibility Signed-off-by: smajumdar <titu1994@gmail.com> * Update training scripts of ASR to support finetuning Signed-off-by: smajumdar <titu1994@gmail.com> * Update Finetuning step to be ModelPT level Signed-off-by: smajumdar <titu1994@gmail.com> * Update docs for finetuning for ASR Signed-off-by: smajumdar <titu1994@gmail.com> * Fix style Signed-off-by: smajumdar <titu1994@gmail.com> * Update docs and scripts with fine-tuning info Signed-off-by: smajumdar <titu1994@gmail.com> * Update docs and scripts with fine-tuning info Signed-off-by: smajumdar <titu1994@gmail.com> * Fix style Signed-off-by: smajumdar <titu1994@gmail.com> * Update scripts Signed-off-by: smajumdar <titu1994@gmail.com> * Add comment for weight initialization Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * TTS Doc Fix and Remove TTS Test (#2272) * bug fix and remove test Signed-off-by: Jason <jasoli@nvidia.com> * syntax Signed-off-by: Jason <jasoli@nvidia.com> * syntax Signed-off-by: Jason <jasoli@nvidia.com> * syntax Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Talknet training Fix (#2273) * TalkNet Training notebook fix. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Remove debug stuff. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update (#2274) Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add links (#2275) * update Signed-off-by: Jason <jasoli@nvidia.com> * link Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Delete 3_TTS_TalkNet_Training.ipynb (#2276) Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * tune down logging (#2277) * tune down logging Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * debug message instead of removing it completely Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * minor bugfix Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * remove confusing message Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Restore TalkNet training notebook (#2281) * Restore TalkNet training notebook. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Remove torchaudio dep. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix ExpManager Issues and FastPitch (#2283) * backport exp_manager fixes to v1 Signed-off-by: Jason <jasoli@nvidia.com> * fix fastpitch Signed-off-by: Jason <jasoli@nvidia.com> * fix tests Signed-off-by: Jason <jasoli@nvidia.com> * update prefix Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Organize asr config folders (#2284) Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix and enable DALI tests (#2077) * Fix and enable DALI tests Signed-off-by: Joaquin Anton <janton@nvidia.com> * remove unused import Signed-off-by: Joaquin Anton <janton@nvidia.com> * Move DALI tests to a separate Jenkins stage Signed-off-by: Joaquin Anton <janton@nvidia.com> * Remove DALI tests from the main jenkins ASR stage Signed-off-by: Joaquin Anton <janton@nvidia.com> * Comment out MFCC test Signed-off-by: Joaquin Anton <janton@nvidia.com> * Working version Signed-off-by: Joaquin Anton <janton@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Added unit test for hifigan export, fixed hifigan export (#2279) * Added unit test for hifigan export, Removed runtime test from waveglow test (now in export) Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixed style Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixed style Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update conformer recipes (#2265) * updated readme asr. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added models. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed the docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed the docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed the docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * disabled test. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Updated the config files. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Updated the config files. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Updated the config files. Signed-off-by: Vahid <vnoroozi@nvidia.com> * dropped the wers. Signed-off-by: Vahid <vnoroozi@nvidia.com> * dropped the wers. Signed-off-by: Vahid <vnoroozi@nvidia.com> * dropped new models and reverted to old versions. Signed-off-by: Vahid <vnoroozi@nvidia.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Adding neural rescorer and its documentations (#2287) * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added more docs, figures, and output file. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added more docs, figures, and output file. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added more docs, figures, and output file. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added more docs, figures, and output file. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added more docs, figures, and output file. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed style Signed-off-by: Vahid <vnoroozi@nvidia.com> * add a note to asr notebook. Signed-off-by: Vahid <vnoroozi@nvidia.com> * add a note to asr notebook. Signed-off-by: Vahid <vnoroozi@nvidia.com> * add a note to asr notebook. Signed-off-by: Vahid <vnoroozi@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Adjust warning messages Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Revert "Adjust warning messages" This reverts commit df046ec55754d0136a2a28451435068f32409f30. Signed-off-by: Micha Livne <mlivne@nvidia.com> * Adjust warning messages (#2294) Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Adding new Models releases on NGC. (#2295) * added new models. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added tests for asr lm. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added tests for asr lm. Signed-off-by: Vahid <vnoroozi@nvidia.com> * dropped the test. Signed-off-by: Vahid <vnoroozi@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update quantization (#2298) Signed-off-by: slyned <slyned@nvidia.com> Co-authored-by: slyned <slyned@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * ASR improvements (#2293) * Update numba messages and citrinet configs Signed-off-by: smajumdar <titu1994@gmail.com> * Remove support for weight init scale and hidden hidden bias scale for layer normalized lstm Signed-off-by: smajumdar <titu1994@gmail.com> * Add support for multiple filetypes in tarred datasets, correct rnn LN-lstm inputs, fix OmegaConf compat issue Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Time quarter to (#2292) * fix comments Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix doc string Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * adding quarter to to time class Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix typo Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fixed paths. (#2301) Signed-off-by: Vahid <vnoroozi@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Added onnxruntime check of exported ONNX, bumped up default ONNX opset (#2278) * Added onnxruntime check of exported ONNX, bumped up default ONNX opset Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Made TS export to accept ONNX-style input example, removed unused param to export Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * check_trace default made False Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixed for updated export signature Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update readmes Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update readme Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update readme Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix docs table Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add support for Numba CUDA optimized SpecAugment (#2269) * Initial implementation Signed-off-by: smajumdar <titu1994@gmail.com> * Initial implementation Signed-off-by: smajumdar <titu1994@gmail.com> * Finish initial implementation of numba spec augment Signed-off-by: smajumdar <titu1994@gmail.com> * Correct mask propagataion Signed-off-by: smajumdar <titu1994@gmail.com> * Parallelize kernel over batch instead of over masks Signed-off-by: smajumdar <titu1994@gmail.com> * Finish tests and update to signature of spectrogramaugmentation calls Signed-off-by: smajumdar <titu1994@gmail.com> * Finish tests and update to signature of spectrogramaugmentation calls Signed-off-by: smajumdar <titu1994@gmail.com> * Add header Signed-off-by: smajumdar <titu1994@gmail.com> * Fix style Signed-off-by: smajumdar <titu1994@gmail.com> * Add heuristics Signed-off-by: smajumdar <titu1994@gmail.com> * Correct inclusive range of padding Signed-off-by: smajumdar <titu1994@gmail.com> * Correct typing for spec aug numba Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Added JSON manifest's support to transcribe_speech.py (#2304) * Added JSON manifest's support to transcribe_speech.py Signed-off-by: Vitaly Lavrukhin <vlavrukhin@nvidia.com> * Dropped unused import Signed-off-by: Vitaly Lavrukhin <vlavrukhin@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * get embedding for a single file (#2310) * get embedding for a single file Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * fixes Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * sr update Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * regain train mode Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update FastPitch (#2249) * wip Signed-off-by: Jason <jasoli@nvidia.com> * c1 Signed-off-by: Jason <jasoli@nvidia.com> * bug fixes Signed-off-by: Jason <jasoli@nvidia.com> * bug fixes Signed-off-by: Jason <jasoli@nvidia.com> * v2 Signed-off-by: Jason <jasoli@nvidia.com> * changes Signed-off-by: Jason <jasoli@nvidia.com> * add types, old model working Signed-off-by: Jason <jasoli@nvidia.com> * pitch Signed-off-by: Jason <jasoli@nvidia.com> * update Signed-off-by: Jason <jasoli@nvidia.com> * update Signed-off-by: Jason <jasoli@nvidia.com> * let it work Signed-off-by: Jason <jasoli@nvidia.com> * fixes Signed-off-by: Jason <jasoli@nvidia.com> * add oktai comments Signed-off-by: Jason <jasoli@nvidia.com> * debug Signed-off-by: Jason <jasoli@nvidia.com> * scale Signed-off-by: Jason <jasoli@nvidia.com> * wip Signed-off-by: Jason <jasoli@nvidia.com> * fix test for v1 Signed-off-by: Jason <jasoli@nvidia.com> * merge train and val Signed-off-by: Jason <jasoli@nvidia.com> * back to par bin att, add correct encoder settings Signed-off-by: Jason <jasoli@nvidia.com> * try Signed-off-by: Jason <jasoli@nvidia.com> * undo Signed-off-by: Jason <jasoli@nvidia.com> * lgtm: Signed-off-by: Jason <jasoli@nvidia.com> * style Signed-off-by: Jason <jasoli@nvidia.com> * default to ljs Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * patch quantization (#2314) * update quantization Signed-off-by: slyned <slyned@nvidia.com> * update quant infer trt Signed-off-by: slyned <slyned@nvidia.com> * fix style Signed-off-by: slyned <slyned@nvidia.com> Co-authored-by: slyned <slyned@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Pin OmegaConf version for 1.0.0 (#2316) * Update OmegaConf compatibility Signed-off-by: smajumdar <titu1994@gmail.com> * Correct OmegaConf.pretty() Signed-off-by: smajumdar <titu1994@gmail.com> * Upper bound omegaconf Signed-off-by: smajumdar <titu1994@gmail.com> * Revert "Correct OmegaConf.pretty()" This reverts commit 6ebae2ef Signed-off-by: smajumdar <titu1994@gmail.com> * Revert "Update OmegaConf compatibility" This reverts commit 83b2cf35a07a742552082e80e6ca34c9b8203cbc. Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * [BUGFIX] OmegaConf forward compatibility (#2319) * Update OmegaConf compatibility Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: ericharper <complex451@gmail.com> * Correct OmegaConf.pretty() Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: ericharper <complex451@gmail.com> * upper bound omegaconf Signed-off-by: ericharper <complex451@gmail.com> * add if,else back Signed-off-by: ericharper <complex451@gmail.com> * typo Signed-off-by: ericharper <complex451@gmail.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * bumping version to 1.0.1 Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix_cluster_small_sample (#2303) * fix_cluster_small_sample Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * for smaller samples Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * remove type Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * similarity matrix Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * est num of speakers add Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * comment update Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * style fix Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * MIN_SAMPLES passed through func arg Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * doc string update Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * spell mistake Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fastpitch export (#2300) * wip Signed-off-by: Jason <jasoli@nvidia.com> * c1 Signed-off-by: Jason <jasoli@nvidia.com> * bug fixes Signed-off-by: Jason <jasoli@nvidia.com> * bug fixes Signed-off-by: Jason <jasoli@nvidia.com> * v2 Signed-off-by: Jason <jasoli@nvidia.com> * changes Signed-off-by: Jason <jasoli@nvidia.com> * add types, old model working Signed-off-by: Jason <jasoli@nvidia.com> * pitch Signed-off-by: Jason <jasoli@nvidia.com> * update Signed-off-by: Jason <jasoli@nvidia.com> * update Signed-off-by: Jason <jasoli@nvidia.com> * let it work Signed-off-by: Jason <jasoli@nvidia.com> * fixes Signed-off-by: Jason <jasoli@nvidia.com> * add oktai comments Signed-off-by: Jason <jasoli@nvidia.com> * debug Signed-off-by: Jason <jasoli@nvidia.com> * scale Signed-off-by: Jason <jasoli@nvidia.com> * wip Signed-off-by: Jason <jasoli@nvidia.com> * fix test for v1 Signed-off-by: Jason <jasoli@nvidia.com> …
fayejf
added a commit
that referenced
this pull request
Jul 16, 2021
* Itn add classes (#2141) * move do_training flag to config Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * added telephone to itn Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * add telephone and email to itn Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * ASR + NLP Doc Fixes (#2136) * Preserve the tokenizer config for ASR Signed-off-by: smajumdar <titu1994@gmail.com> * Correct nlp docs Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Removing graphsurgeon optional dependency, improving import error rep… (#2144) * Removing graphsurgeon optional dependency, improving import error reporting Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixing scope error Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix FilterbankFeatures eval nondeterminism. (#2146) Signed-off-by: PiotrDabkowski <piodrus@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix the docs. (#2148) Signed-off-by: Micha Livne <mlivne@nvidia.com> * Text processing refactor (#2149) * removed graphutils, suppletive, data_loader_utils from itn to be reused from tn Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * inheriting itn from tn, thus removing redundancy Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * cleaned whitelist Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * lgtm fix Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update how artifacts work (#2138) * Update how artifacts work Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * fixing some tests Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * fix more tests Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * add __init__ to tests to make them discoverable Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * empty src support Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * updates plust unittest Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * add copyright check Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * copyright header Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * fix style Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * handle hashed megatron checkpoint version in nlp restore_from Signed-off-by: ericharper <complex451@gmail.com> * add _MODEL_RESTORE_PATH to AppState Signed-off-by: ericharper <complex451@gmail.com> * get rid of global folder caching Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * double register - warning instead of exception Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * Add asr spe tests Signed-off-by: smajumdar <titu1994@gmail.com> * Pop out asr wpe pre-registered value Signed-off-by: smajumdar <titu1994@gmail.com> * Correct ASR tests and paths Signed-off-by: smajumdar <titu1994@gmail.com> * Correct tokenizer saving Signed-off-by: smajumdar <titu1994@gmail.com> * Correct ASR tests Signed-off-by: smajumdar <titu1994@gmail.com> * Correct ASR bpe mixin Signed-off-by: smajumdar <titu1994@gmail.com> * Patch up backward compatibility Signed-off-by: smajumdar <titu1994@gmail.com> * update register_bert_model Signed-off-by: ericharper <complex451@gmail.com> * update all get_lm_model calls Signed-off-by: ericharper <complex451@gmail.com> * return None if src not found Signed-off-by: ericharper <complex451@gmail.com> * handle case with no tokenizer Signed-off-by: ericharper <complex451@gmail.com> * do not add another hash is using tarfile_artifacts Signed-off-by: ericharper <complex451@gmail.com> * add return_none flag, update doc string Signed-off-by: ericharper <complex451@gmail.com> * update default behavior of register_artifact for NLPModel Signed-off-by: ericharper <complex451@gmail.com> * change kwarg name to verify_src_exists Signed-off-by: ericharper <complex451@gmail.com> * use cfg instead of _cfg Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * some cleanups Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Co-authored-by: ericharper <complex451@gmail.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Language model refactoring (#2120) * fixed branch in IR tutorial Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * bucketing tarred dataset for lm training Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * updated global rank Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * perplexity update Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * refactor lm to be campatible with latest nmt Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * perplexity change Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * removed obsolete config Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * added sequence perplexity Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * added non-smoothed CE loss for validation Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * unified sentence dataset, torchmetrics for sequence perplexity Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * translate_ddp refactor Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Signed-off-by: Micha Livne <mlivne@nvidia.com> * [NMT] Multi-validation Patch (#2150) * rename dl index 0 loss and sacrebleu for backwards compatibility Signed-off-by: ericharper <complex451@gmail.com> * eval -> val/tst Signed-off-by: ericharper <complex451@gmail.com> * instantiate torchmetrics after instantiating dataloaders Signed-off-by: ericharper <complex451@gmail.com> * bug Signed-off-by: ericharper <complex451@gmail.com> * remove debugging log Signed-off-by: ericharper <complex451@gmail.com> * remove debugging log Signed-off-by: ericharper <complex451@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * bumping version to 1.0.0 Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fixed the num_samples of text classification model. (#2152) Signed-off-by: Vahid <vnoroozi@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix for electronic (#2153) * fix for electronic Signed-off-by: ekmb <ebakhturina@nvidia.com> * special symbols added Signed-off-by: ekmb <ebakhturina@nvidia.com> * restrict symbols list Signed-off-by: ekmb <ebakhturina@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * FastSpeech 2 Test & Docs (#2143) * Add FS2 data loading test Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * TTS docs update for FastSpeech 2 Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Style fix for FS2 dataset test Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Fix transpose typo Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Minor patch for translate_ddp (#2155) * Patch for backtranslation in lm dataset Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * One more fix Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Entity linking (#2050) * Started adding SAP dataset Signed-off-by: Virginia Adams <vadams@nvidia.com> * Delete .lm_bert_dataset.py.swp Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added dataset and loss Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added entity linking encoder model Signed-off-by: Virginia Adams <vadams@nvidia.com> * Can build and use index from pubmedbert model Signed-off-by: Virginia Adams <vadams@nvidia.com> * checked boolean logic in build_index.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * End to end tested all functionality Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed val loss none at end of validation Signed-off-by: Virginia Adams <vadams@nvidia.com> * Started adding demo entity linking notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * adding in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * added call to entitylinking classes in __init__.py files Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added eval code to notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * Adding unfinished notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * Cleaned up example dir Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed recap commands Signed-off-by: Virginia Adams <vadams@nvidia.com> * added model typing and tiny data tar Signed-off-by: Virginia Adams <vadams@nvidia.com> * Adding tiny data zip Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated tiny example config data path Signed-off-by: Virginia Adams <vadams@nvidia.com> * Notebook demo works Signed-off-by: Virginia Adams <vadams@nvidia.com> * Changed training epochs Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed output from training and install cells Signed-off-by: Virginia Adams <vadams@nvidia.com> * changed code formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Started doc string for new functions Signed-off-by: Virginia Adams <vadams@nvidia.com> * Updated data_preprocessing to save to data_dir Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed comment in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * Update data_preprocessing.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated nemo typing imports Signed-off-by: Virginia Adams <vadams@nvidia.com> * about to rebase Signed-off-by: Virginia Adams <vadams@nvidia.com> * added back umls_dataset_processing.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed example data Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed typos in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed lgtm-com issues Signed-off-by: Virginia Adams <vadams@nvidia.com> * added copyright headers Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed import and copyright headers Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed formatting changes 2 Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed test formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added __init__.py for model and dataset Signed-off-by: Virginia Adams <vadams@nvidia.com> * loading newline file returns data_dir now Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed conf notebook and deleted comment Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * Updated Jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed file path Signed-off-by: Virginia Adams <vadams@nvidia.com> * Changed Jenkins pipeline order Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed Jenkins datapath... again... Signed-off-by: Virginia Adams <vadams@nvidia.com> * Made most review changes Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed copy right Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated unit test to wget config Signed-off-by: Virginia Adams <vadams@nvidia.com> * reverted test file back Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added project dir to jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * defined config in unit test Signed-off-by: Virginia Adams <vadams@nvidia.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Correct branch version for v1.0.0 (#2157) * Correct branch version Signed-off-by: smajumdar <titu1994@gmail.com> * Correct Jenkinsfile Signed-off-by: smajumdar <titu1994@gmail.com> * Update rst files Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * switch CI back to main Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fixed the docs. (#2156) Signed-off-by: Micha Livne <mlivne@nvidia.com> * Make Hifigan jittable (#2159) * FastSpeech 2 Test & Docs (#2143) * Add FS2 data loading test Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * TTS docs update for FastSpeech 2 Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Style fix for FS2 dataset test Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Fix transpose typo Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Jason <jasoli@nvidia.com> * Entity linking (#2050) * Started adding SAP dataset Signed-off-by: Virginia Adams <vadams@nvidia.com> * Delete .lm_bert_dataset.py.swp Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added dataset and loss Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added entity linking encoder model Signed-off-by: Virginia Adams <vadams@nvidia.com> * Can build and use index from pubmedbert model Signed-off-by: Virginia Adams <vadams@nvidia.com> * checked boolean logic in build_index.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * End to end tested all functionality Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed val loss none at end of validation Signed-off-by: Virginia Adams <vadams@nvidia.com> * Started adding demo entity linking notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * adding in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * added call to entitylinking classes in __init__.py files Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added eval code to notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * Adding unfinished notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * Cleaned up example dir Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed recap commands Signed-off-by: Virginia Adams <vadams@nvidia.com> * added model typing and tiny data tar Signed-off-by: Virginia Adams <vadams@nvidia.com> * Adding tiny data zip Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated tiny example config data path Signed-off-by: Virginia Adams <vadams@nvidia.com> * Notebook demo works Signed-off-by: Virginia Adams <vadams@nvidia.com> * Changed training epochs Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed output from training and install cells Signed-off-by: Virginia Adams <vadams@nvidia.com> * changed code formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Started doc string for new functions Signed-off-by: Virginia Adams <vadams@nvidia.com> * Updated data_preprocessing to save to data_dir Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed comment in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * Update data_preprocessing.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated nemo typing imports Signed-off-by: Virginia Adams <vadams@nvidia.com> * about to rebase Signed-off-by: Virginia Adams <vadams@nvidia.com> * added back umls_dataset_processing.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed example data Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed typos in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed lgtm-com issues Signed-off-by: Virginia Adams <vadams@nvidia.com> * added copyright headers Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed import and copyright headers Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed formatting changes 2 Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed test formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added __init__.py for model and dataset Signed-off-by: Virginia Adams <vadams@nvidia.com> * loading newline file returns data_dir now Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed conf notebook and deleted comment Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * Updated Jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed file path Signed-off-by: Virginia Adams <vadams@nvidia.com> * Changed Jenkins pipeline order Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed Jenkins datapath... again... Signed-off-by: Virginia Adams <vadams@nvidia.com> * Made most review changes Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed copy right Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated unit test to wget config Signed-off-by: Virginia Adams <vadams@nvidia.com> * reverted test file back Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added project dir to jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * defined config in unit test Signed-off-by: Virginia Adams <vadams@nvidia.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> * switch CI back to main Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * Make Hifigan jittable Signed-off-by: Ryan Leary <rleary@nvidia.com> * Remove vestigial debugging printout Signed-off-by: Ryan Leary <rleary@nvidia.com> * Add export forward and fix style Signed-off-by: Ryan Leary <rleary@nvidia.com> * Fix load_state_dict override for arbitrary layers Signed-off-by: Ryan Leary <rleary@nvidia.com> Co-authored-by: Jocelyn <jocelynh@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Jason <jasoli@nvidia.com> Co-authored-by: vadam5 <78445382+vadam5@users.noreply.github.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Co-authored-by: Ryan Leary <rleary@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix version (#2162) Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Megatron nb size reduced (#2163) * notebook size reduced Signed-off-by: ekmb <ebakhturina@nvidia.com> * notebook size reduced Signed-off-by: ekmb <ebakhturina@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update spectral clustering method (#2158) * update spectral clustering method Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * update Jenkins File Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * threshold fix by reducing window length for shorter embs Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * grammar fixes Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * CR update Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * paper reference Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * improve docstring for yaml Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * Doc fixes Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * revert (#2167) Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Limit Pytorch lightning release (#2170) Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * token classification models artifacts update (#2169) * artifacts update Signed-off-by: ekmb <ebakhturina@nvidia.com> * artifacts update Signed-off-by: ekmb <ebakhturina@nvidia.com> * jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> * fix for model restoration Signed-off-by: ekmb <ebakhturina@nvidia.com> * typos fix + jenkins dir update Signed-off-by: ekmb <ebakhturina@nvidia.com> * typo Signed-off-by: ericharper <complex451@gmail.com> * update jenkins branch Signed-off-by: ericharper <complex451@gmail.com> * add && Signed-off-by: ericharper <complex451@gmail.com> * jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> * jenkins disable Signed-off-by: ekmb <ebakhturina@nvidia.com> * revert jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> * jenkins disable Signed-off-by: ekmb <ebakhturina@nvidia.com> * revert jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> Co-authored-by: ericharper <complex451@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix to always_save_nemo (#2174) * Initial attempt at always_save_nemo fix Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * updated path before saving in exp manager, fixed bug when handling tarfile artifacts Signed-off-by: ericharper <complex451@gmail.com> * Add test with always_save_nemo to exp_manager Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * Style fixes Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * update jenkins branch Signed-off-by: ericharper <complex451@gmail.com> * check for nemo: Signed-off-by: ericharper <complex451@gmail.com> * check for nemo: Signed-off-by: ericharper <complex451@gmail.com> * check for nemo: Signed-off-by: ericharper <complex451@gmail.com> Co-authored-by: ericharper <complex451@gmail.com> Co-authored-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix typo (#2179) Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Make itn tests optional (#2173) * Limit Pytorch lightning release Signed-off-by: smajumdar <titu1994@gmail.com> * Add final two checks Signed-off-by: smajumdar <titu1994@gmail.com> Co-authored-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * First Revision of TTS Docs and Notebooks Update for 1.0 (#2166) * squash Signed-off-by: Jason <jasoli@nvidia.com> * notebook fixes Signed-off-by: Jason <jasoli@nvidia.com> * notebook fixes Signed-off-by: Jason <jasoli@nvidia.com> * typos Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * add more alternatives of 0 for telephone (#2171) Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Acc tn (#2180) * make tn cardinal faster Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * add number far Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * add test Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix lgtm Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix lgtm Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * [DOCS] NLP Model parallel, NMT multi-val, CORE register artifacts (#2168) * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Change label smoothing prob to reduce chance of test failure (#2184) Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Eric Harper <complex451@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add FS2 checkpoint links to docs and inference notebook (#2181) * Add FS2 checkpoint links to docs and inference notebook Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Remove empty cell from TTS notebook Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update ptl to 1.3 on main branch (#2178) * Update PTL Signed-off-by: smajumdar <titu1994@gmail.com> * Begin update to Pytorch Lightning 1.3.x Signed-off-by: smajumdar <titu1994@gmail.com> * Formatting Signed-off-by: smajumdar <titu1994@gmail.com> * style Signed-off-by: ericharper <complex451@gmail.com> * Formatting Signed-off-by: smajumdar <titu1994@gmail.com> * minor fix Signed-off-by: Jason <jasoli@nvidia.com> * minor fix Signed-off-by: Jason <jasoli@nvidia.com> * get testing attribute from trainer Signed-off-by: ericharper <complex451@gmail.com> * update init_ddp_connection override Signed-off-by: ericharper <complex451@gmail.com> * update attribute Signed-off-by: ericharper <complex451@gmail.com> * add barrier after load checkpoint in megatron Signed-off-by: ericharper <complex451@gmail.com> * remove barrier Signed-off-by: ericharper <complex451@gmail.com> * update last naming Signed-off-by: Jason <jasoli@nvidia.com> Co-authored-by: ericharper <complex451@gmail.com> Co-authored-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * SDE updates (#2187) * Added updates to SDE: - support for external vocabulary (to detect OOV words) - support for offset field (for segmented long recordings) - UI improvements Signed-off-by: Vitaly Lavrukhin <vlavrukhin@nvidia.com> * Refactored diff in SDE Signed-off-by: Vitaly Lavrukhin <vlavrukhin@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add TTS aligner and improved version of g2p for vocabs.Phonemes, small improvement in TalkNet (#2189) * add first version of aligner Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> * aligner docs, new g2p version, fix bugs in talknet Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> * update docs and remove lj related code Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> * fix style Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> * fix import Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * set the default of nodessplitter to None. (#2190) Signed-off-by: Vahid <vnoroozi@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * NMT fixes (#2194) * minor fixes Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * minor bugfixes Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Store mappings file in .nemo for FS2 model (#2196) * Store mappings file in .nemo for FS2 model Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Add error enforcing mappings file during training (FS2) Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add support to change the SE context window of ConvASREncoder (#2193) * Add support for changing context window on the fly Signed-off-by: smajumdar <titu1994@gmail.com> * Add support to change the SE context window of ConvASREncoder Signed-off-by: smajumdar <titu1994@gmail.com> * Add ability to skip config updating Signed-off-by: smajumdar <titu1994@gmail.com> * Switch to mixin based API Signed-off-by: smajumdar <titu1994@gmail.com> * Update docs and api for ASRModuleMixin Signed-off-by: smajumdar <titu1994@gmail.com> * Change print to logging.info Signed-off-by: smajumdar <titu1994@gmail.com> * Correct stride level when computing context window Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add a CI test for doing inference with an NMT model trained with Pre-LN (#2198) * Change label smoothing prob to reduce chance of test failure Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * Add Pre-LN inference test to Jenkinsfile Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * Separate tests for training and NMT inference Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix ipywidgets error in asr notebook (#2199) Added `ipywidgets` to avoid `ImportError: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html` error. Signed-off-by: Derek Chia <derek@derekchia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * metrics fix (#2202) * metrics fix Signed-off-by: ekmb <ebakhturina@nvidia.com> * metrics reset for punct model Signed-off-by: ekmb <ebakhturina@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * readme and minor improvements (#2203) * readme and minor improvements Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * vad threshold update Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix text processing docs (#2195) * fix text processing docs Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix name Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * add guard to pynini import Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix bug in SpecCutout (#2201) Signed-off-by: Robert Bracco <robertbracco1@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix bug in SpecCutout (#2201) (#2205) Signed-off-by: Robert Bracco <robertbracco1@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: smajumdar <titu1994@gmail.com> Co-authored-by: Robert Bracco <47190785+rbracco@users.noreply.github.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Set seed before generating random tensors in NMT test (#2206) * Change label smoothing prob to reduce chance of test failure Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * Set seed before generating tensors Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> Signed-off-by: Micha Livne <mlivne@nvidia.com> * ASR patches for v1.0.0 (#2207) * Multiple updates to RNNT add initialization Signed-off-by: smajumdar <titu1994@gmail.com> * Correct name of initilization Signed-off-by: smajumdar <titu1994@gmail.com> * Update dockerignore Signed-off-by: smajumdar <titu1994@gmail.com> * Fix RNNT WER calculation Signed-off-by: smajumdar <titu1994@gmail.com> * Address comments Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Multilingual training for NMT (#2160) * mnmt on fresh main Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * push for test Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * debug Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * check Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * cleanup Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * minor fix Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * more minor fixes Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * fix for test Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * fix list size error Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * multilingual in infer Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * changes Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * tar creation with multilingual Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * fix Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * changes + parallelism + bug fix Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * small fix Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * multilingual preprocessor fix Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * globally unique fragment names in tarred dataset Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * minor changes Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * rm load_from_cached_dataset Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * minor config change Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * rm unsued import Signed-off-by: Abhinav Khattar <aklife97@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Remove memory leak from ASR notebook + update model notebook (#2213) * ASR patches for v1.0.0 (#2207) * Multiple updates to RNNT add initialization Signed-off-by: smajumdar <titu1994@gmail.com> * Correct name of initilization Signed-off-by: smajumdar <titu1994@gmail.com> * Update dockerignore Signed-off-by: smajumdar <titu1994@gmail.com> * Fix RNNT WER calculation Signed-off-by: smajumdar <titu1994@gmail.com> * Address comments Signed-off-by: smajumdar <titu1994@gmail.com> * Correct model notebook to log the loss and correctly assign keys Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * replace names in vad tutorials (#2220) Signed-off-by: fayejf <fayejf07@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix the versioning name. (#2209) * fix the versioning name. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Made version None. Signed-off-by: Vahid <vnoroozi@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Enabled passing kwargs to export() (#2175) * Enabled passing kwargs to export() Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixing style; changed Classifier input_example to new extended syntax Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixed order of forward() call in export Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixing style Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update g2p: ambigious ignore, flag for skipping seq2seq (#2223) Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update TTS notebook with TalkNet inference (#2133) * Update TTS notebook with TalkNet inference. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Update TTS Notebook. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Update TTS TN Training Notebook. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Fix TN paper link. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Remove branch updaing TODOs. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update speaker notebooks (#2224) Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Support symlinked files (#2216) Signed-off-by: Anas Abou Allaban <aabouallaban@pm.me> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Set strict=True everywhere by default. (#2225) Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * set strict=True in nlp_model (#2227) Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * set strict=False for model parallel examples Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Make Text processing installation optional via reinstall.sh (#2226) * Make Text processing installation optional via reinstall.sh Signed-off-by: smajumdar <titu1994@gmail.com> * Support both success and failure states Signed-off-by: smajumdar <titu1994@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Transformer final norm preln (#2197) * fix pre_ln final norm Signed-off-by: fayejf <fayejf07@gmail.com> * style fix Signed-off-by: fayejf <fayejf07@gmail.com> * bug fixed Signed-off-by: fayejf <fayejf07@gmail.com> * bugfix post_ln Signed-off-by: fayejf <fayejf07@gmail.com> * update and add pre_ln_final_norm Signed-off-by: fayejf <fayejf07@gmail.com> * style fix Signed-off-by: fayejf <fayejf07@gmail.com> * fix for unit test Signed-off-by: fayejf <fayejf07@gmail.com> * rename final_norm to final_layer_norm Signed-off-by: fayejf <fayejf07@gmail.com> * bug fix Signed-off-by: fayejf <fayejf07@gmail.com> * tiny fix Signed-off-by: fayejf <fayejf07@gmail.com> * fix and improve Signed-off-by: fayejf <fayejf07@gmail.com> * tiny fix Signed-off-by: fayejf <fayejf07@gmail.com> * Patch for NMT to allow loading old modlels trained with pre-LN Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update models and notebook for 1.0 (#2211) * update models Signed-off-by: Jason <jasoli@nvidia.com> * updates Signed-off-by: Jason <jasoli@nvidia.com> * fix Signed-off-by: Jason <jasoli@nvidia.com> * add links Signed-off-by: Jason <jasoli@nvidia.com> * fixes Signed-off-by: Jason <jasoli@nvidia.com> * style Signed-off-by: Jason <jasoli@nvidia.com> * update checkpoints Signed-off-by: Jason <jasoli@nvidia.com> * rename Signed-off-by: Jason <jasoli@nvidia.com> * fixes Signed-off-by: Jason <jasoli@nvidia.com> * lgtm Signed-off-by: Jason <jasoli@nvidia.com> * fix loading waveglow Signed-off-by: Jason <jasoli@nvidia.com> * typo Signed-off-by: Jason <jasoli@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update_metrics_classification_models (#2228) Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Co-authored-by: fayejf <36722593+fayejf@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Data loader for seq of label model (#2084) * feature to seq label data loader Signed-off-by: fayejf <fayejf07@gmail.com> * style fix Signed-off-by: fayejf <fayejf07@gmail.com> * small fix Signed-off-by: fayejf <fayejf07@gmail.com> * update tl to be length of seq label Signed-off-by: fayejf <fayejf07@gmail.com> * style fix Signed-off-by: fayejf <fayejf07@gmail.com> * tiny bug fix Signed-off-by: fayejf <fayejf07@gmail.com> * small updates Signed-off-by: fayejf <fayejf07@gmail.com> * updates for review feedback Signed-off-by: fayejf <fayejf07@gmail.com> * style fix Signed-off-by: fayejf <fayejf07@gmail.com> * explain seq_label Signed-off-by: fayejf <fayejf07@gmail.com> * fix lgtm Signed-off-by: fayejf <fayejf07@gmail.com> * small updates Signed-off-by: fayejf <fayejf07@gmail.com> * improve as discussed Signed-off-by: fayejf <fayejf07@gmail.com> * add docstring Signed-off-by: fayejf <fayejf07@gmail.com> Co-authored-by: Nithin Rao <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix comments (#2236) * fix comments Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix typo Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * add paper ref to sgdqa model doc (#2233) * add paper ref to sgdqa model doc Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix comments Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Move ConcatDataset to common (#2237) * move concatdataset to common Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * var name change Signed-off-by: Abhinav Khattar <aklife97@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * audio based normalization (#2231) * squash norm_audio Signed-off-by: ekmb <ebakhturina@nvidia.com> * add missing files Signed-off-by: ekmb <ebakhturina@nvidia.com> * style Signed-off-by: ekmb <ebakhturina@nvidia.com> * unit tests added, docstrings fixed Signed-off-by: ekmb <ebakhturina@nvidia.com> * fix lgtm errors Signed-off-by: ekmb <ebakhturina@nvidia.com> * debug jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> * debug jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> * signature update Signed-off-by: ekmb <ebakhturina@nvidia.com> * set deterministic default Signed-off-by: ekmb <ebakhturina@nvidia.com> * add more test cases Signed-off-by: ekmb <ebakhturina@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * bug fix config (#2232) Signed-off-by: fayejf <fayejf07@gmail.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Alias Swish to SiLU (#2239) * Alias Swish to SiLU and move activations to inplace execution if possible Signed-off-by: smajumdar <titu1994@gmail.com> * Remove unused import Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update README.rst Signed-off-by: Micha Livne <mlivne@nvidia.com> * Offline asr notebook bug fix (#2242) * fix Signed-off-by: fayejf <fayejf07@gmail.com> * install Signed-off-by: fayejf <fayejf07@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix docstring (#2244) * fix comments Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix doc string Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update "last" Checkpoint (#2241) * fix Signed-off-by: Jason <jasoli@nvidia.com> * change Signed-off-by: Jason <jasoli@nvidia.com> * fix Signed-off-by: Jason <jasoli@nvidia.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add pretrained model stt_es_citrinet_512 (#2247) Signed-off-by: Elena Rastorgueva <erastorgueva@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * [BUGFIX] Only process tarfile artifacts when model was restored from tarfile (#2250) * process tarfile artifacts only if model is being restored Signed-off-by: ericharper <complex451@gmail.com> * process tarfile artifacts only if model was restored from a tarfile Signed-off-by: ericharper <complex451@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Log average metrics for Multi-validation in NMT (#2251) * add avg metrics NMT Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * name change Signed-off-by: Abhinav Khattar <aklife97@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update Primer notebook (#2258) Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fixed Bug 3310780 and 3310799 (#2264) Signed-off-by: Virginia Adams <vadams@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Support multiple models being instantiated in same execution scope (#2245) * Support multiple models being instantiated in same execution scope Signed-off-by: smajumdar <titu1994@gmail.com> * Fix tests Signed-off-by: smajumdar <titu1994@gmail.com> * Add locks to methods in appstate Signed-off-by: smajumdar <titu1994@gmail.com> * Perform locks only on write operations Signed-off-by: smajumdar <titu1994@gmail.com> * Correct deadlock issue Signed-off-by: smajumdar <titu1994@gmail.com> * Add more tests Signed-off-by: smajumdar <titu1994@gmail.com> * Add test for multi save and remove patch to change save type Signed-off-by: smajumdar <titu1994@gmail.com> * Update app state to preserve gidx of previous token Signed-off-by: smajumdar <titu1994@gmail.com> * Correct restoration logic for tarfiles Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * ASR Refactoring (#2240) * Refactor out the preprocessing from ASR into common Signed-off-by: smajumdar <titu1994@gmail.com> * Correct nltk issue with vocabs.py for clusters Signed-off-by: smajumdar <titu1994@gmail.com> * Add typing information to SpecAugment and SpecCutout Signed-off-by: smajumdar <titu1994@gmail.com> * Reorganize parts directory Signed-off-by: smajumdar <titu1994@gmail.com> * Refactor parts submodules, add __init__ to few important parts Signed-off-by: smajumdar <titu1994@gmail.com> * Update docs for new path to parts Signed-off-by: smajumdar <titu1994@gmail.com> * Cherry pick PR https://github.com/NVIDIA/NeMo/pull/2219 Signed-off-by: smajumdar <titu1994@gmail.com> * Add header for preprocessing commons Signed-off-by: smajumdar <titu1994@gmail.com> * Fix style of tests Signed-off-by: smajumdar <titu1994@gmail.com> * Add forced update of configs for train-val-test ds to new labels tests Signed-off-by: smajumdar <titu1994@gmail.com> * Update path to FilterbankFeatures for TTS Signed-off-by: smajumdar <titu1994@gmail.com> * Add an alias file for backward compatibility Signed-off-by: smajumdar <titu1994@gmail.com> * Add an alias file for backward compatibility Signed-off-by: smajumdar <titu1994@gmail.com> * Update training scripts of ASR to support finetuning Signed-off-by: smajumdar <titu1994@gmail.com> * Update Finetuning step to be ModelPT level Signed-off-by: smajumdar <titu1994@gmail.com> * Update docs for finetuning for ASR Signed-off-by: smajumdar <titu1994@gmail.com> * Fix style Signed-off-by: smajumdar <titu1994@gmail.com> * Update docs and scripts with fine-tuning info Signed-off-by: smajumdar <titu1994@gmail.com> * Update docs and scripts with fine-tuning info Signed-off-by: smajumdar <titu1994@gmail.com> * Fix style Signed-off-by: smajumdar <titu1994@gmail.com> * Update scripts Signed-off-by: smajumdar <titu1994@gmail.com> * Add comment for weight initialization Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * TTS Doc Fix and Remove TTS Test (#2272) * bug fix and remove test Signed-off-by: Jason <jasoli@nvidia.com> * syntax Signed-off-by: Jason <jasoli@nvidia.com> * syntax Signed-off-by: Jason <jasoli@nvidia.com> * syntax Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Talknet training Fix (#2273) * TalkNet Training notebook fix. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Remove debug stuff. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update (#2274) Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add links (#2275) * update Signed-off-by: Jason <jasoli@nvidia.com> * link Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Delete 3_TTS_TalkNet_Training.ipynb (#2276) Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * tune down logging (#2277) * tune down logging Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * debug message instead of removing it completely Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * minor bugfix Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * remove confusing message Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Restore TalkNet training notebook (#2281) * Restore TalkNet training notebook. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Remove torchaudio dep. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix ExpManager Issues and FastPitch (#2283) * backport exp_manager fixes to v1 Signed-off-by: Jason <jasoli@nvidia.com> * fix fastpitch Signed-off-by: Jason <jasoli@nvidia.com> * fix tests Signed-off-by: Jason <jasoli@nvidia.com> * update prefix Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Organize asr config folders (#2284) Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix and enable DALI tests (#2077) * Fix and enable DALI tests Signed-off-by: Joaquin Anton <janton@nvidia.com> * remove unused import Signed-off-by: Joaquin Anton <janton@nvidia.com> * Move DALI tests to a separate Jenkins stage Signed-off-by: Joaquin Anton <janton@nvidia.com> * Remove DALI tests from the main jenkins ASR stage Signed-off-by: Joaquin Anton <janton@nvidia.com> * Comment out MFCC test Signed-off-by: Joaquin Anton <janton@nvidia.com> * Working version Signed-off-by: Joaquin Anton <janton@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Added unit test for hifigan export, fixed hifigan export (#2279) * Added unit test for hifigan export, Removed runtime test from waveglow test (now in export) Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixed style Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixed style Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update conformer recipes (#2265) * updated readme asr. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added models. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed the docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed the docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed the docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * disabled test. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Updated the config files. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Updated the config files. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Updated the config files. Signed-off-by: Vahid <vnoroozi@nvidia.com> * dropped the wers. Signed-off-by: Vahid <vnoroozi@nvidia.com> * dropped the wers. Signed-off-by: Vahid <vnoroozi@nvidia.com> * dropped new models and reverted to old versions. Signed-off-by: Vahid <vnoroozi@nvidia.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Adding neural rescorer and its documentations (#2287) * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added more docs, figures, and output file. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added more docs, figures, and output file. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added more docs, figures, and output file. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added more docs, figures, and output file. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added more docs, figures, and output file. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed style Signed-off-by: Vahid <vnoroozi@nvidia.com> * add a note to asr notebook. Signed-off-by: Vahid <vnoroozi@nvidia.com> * add a note to asr notebook. Signed-off-by: Vahid <vnoroozi@nvidia.com> * add a note to asr notebook. Signed-off-by: Vahid <vnoroozi@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Adjust warning messages Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Revert "Adjust warning messages" This reverts commit df046ec55754d0136a2a28451435068f32409f30. Signed-off-by: Micha Livne <mlivne@nvidia.com> * Adjust warning messages (#2294) Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Adding new Models releases on NGC. (#2295) * added new models. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added tests for asr lm. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added tests for asr lm. Signed-off-by: Vahid <vnoroozi@nvidia.com> * dropped the test. Signed-off-by: Vahid <vnoroozi@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update quantization (#2298) Signed-off-by: slyned <slyned@nvidia.com> Co-authored-by: slyned <slyned@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * ASR improvements (#2293) * Update numba messages and citrinet configs Signed-off-by: smajumdar <titu1994@gmail.com> * Remove support for weight init scale and hidden hidden bias scale for layer normalized lstm Signed-off-by: smajumdar <titu1994@gmail.com> * Add support for multiple filetypes in tarred datasets, correct rnn LN-lstm inputs, fix OmegaConf compat issue Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Time quarter to (#2292) * fix comments Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix doc string Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * adding quarter to to time class Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix typo Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fixed paths. (#2301) Signed-off-by: Vahid <vnoroozi@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Added onnxruntime check of exported ONNX, bumped up default ONNX opset (#2278) * Added onnxruntime check of exported ONNX, bumped up default ONNX opset Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Made TS export to accept ONNX-style input example, removed unused param to export Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * check_trace default made False Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixed for updated export signature Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update readmes Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update readme Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update readme Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix docs table Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add support for Numba CUDA optimized SpecAugment (#2269) * Initial implementation Signed-off-by: smajumdar <titu1994@gmail.com> * Initial implementation Signed-off-by: smajumdar <titu1994@gmail.com> * Finish initial implementation of numba spec augment Signed-off-by: smajumdar <titu1994@gmail.com> * Correct mask propagataion Signed-off-by: smajumdar <titu1994@gmail.com> * Parallelize kernel over batch instead of over masks Signed-off-by: smajumdar <titu1994@gmail.com> * Finish tests and update to signature of spectrogramaugmentation calls Signed-off-by: smajumdar <titu1994@gmail.com> * Finish tests and update to signature of spectrogramaugmentation calls Signed-off-by: smajumdar <titu1994@gmail.com> * Add header Signed-off-by: smajumdar <titu1994@gmail.com> * Fix style Signed-off-by: smajumdar <titu1994@gmail.com> * Add heuristics Signed-off-by: smajumdar <titu1994@gmail.com> * Correct inclusive range of padding Signed-off-by: smajumdar <titu1994@gmail.com> * Correct typing for spec aug numba Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Added JSON manifest's support to transcribe_speech.py (#2304) * Added JSON manifest's support to transcribe_speech.py Signed-off-by: Vitaly Lavrukhin <vlavrukhin@nvidia.com> * Dropped unused import Signed-off-by: Vitaly Lavrukhin <vlavrukhin@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * get embedding for a single file (#2310) * get embedding for a single file Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * fixes Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * sr update Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * regain train mode Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update FastPitch (#2249) * wip Signed-off-by: Jason <jasoli@nvidia.com> * c1 Signed-off-by: Jason <jasoli@nvidia.com> * bug fixes Signed-off-by: Jason <jasoli@nvidia.com> * bug fixes Signed-off-by: Jason <jasoli@nvidia.com> * v2 Signed-off-by: Jason <jasoli@nvidia.com> * changes Signed-off-by: Jason <jasoli@nvidia.com> * add types, old model working Signed-off-by: Jason <jasoli@nvidia.com> * pitch Signed-off-by: Jason <jasoli@nvidia.com> * update Signed-off-by: Jason <jasoli@nvidia.com> * update Signed-off-by: Jason <jasoli@nvidia.com> * let it work Signed-off-by: Jason <jasoli@nvidia.com> * fixes Signed-off-by: Jason <jasoli@nvidia.com> * add oktai comments Signed-off-by: Jason <jasoli@nvidia.com> * debug Signed-off-by: Jason <jasoli@nvidia.com> * scale Signed-off-by: Jason <jasoli@nvidia.com> * wip Signed-off-by: Jason <jasoli@nvidia.com> * fix test for v1 Signed-off-by: Jason <jasoli@nvidia.com> * merge train and val Signed-off-by: Jason <jasoli@nvidia.com> * back to par bin att, add correct encoder settings Signed-off-by: Jason <jasoli@nvidia.com> * try Signed-off-by: Jason <jasoli@nvidia.com> * undo Signed-off-by: Jason <jasoli@nvidia.com> * lgtm: Signed-off-by: Jason <jasoli@nvidia.com> * style Signed-off-by: Jason <jasoli@nvidia.com> * default to ljs Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * patch quantization (#2314) * update quantization Signed-off-by: slyned <slyned@nvidia.com> * update quant infer trt Signed-off-by: slyned <slyned@nvidia.com> * fix style Signed-off-by: slyned <slyned@nvidia.com> Co-authored-by: slyned <slyned@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Pin OmegaConf version for 1.0.0 (#2316) * Update OmegaConf compatibility Signed-off-by: smajumdar <titu1994@gmail.com> * Correct OmegaConf.pretty() Signed-off-by: smajumdar <titu1994@gmail.com> * Upper bound omegaconf Signed-off-by: smajumdar <titu1994@gmail.com> * Revert "Correct OmegaConf.pretty()" This reverts commit 6ebae2ef Signed-off-by: smajumdar <titu1994@gmail.com> * Revert "Update OmegaConf compatibility" This reverts commit 83b2cf35a07a742552082e80e6ca34c9b8203cbc. Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * [BUGFIX] OmegaConf forward compatibility (#2319) * Update OmegaConf compatibility Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: ericharper <complex451@gmail.com> * Correct OmegaConf.pretty() Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: ericharper <complex451@gmail.com> * upper bound omegaconf Signed-off-by: ericharper <complex451@gmail.com> * add if,else back Signed-off-by: ericharper <complex451@gmail.com> * typo Signed-off-by: ericharper <complex451@gmail.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * bumping version to 1.0.1 Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix_cluster_small_sample (#2303) * fix_cluster_small_sample Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * for smaller samples Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * remove type Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * similarity matrix Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * est num of speakers add Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * comment update Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * style fix Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * MIN_SAMPLES passed through func arg Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * doc string update Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * spell mistake Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fastpitch export (#2300) * wip Signed-off-by: Jason <jasoli@nvidia.com> * c1 Signed-off-by: Jason <jasoli@nvidia.com> * bug fixes Signed-off-by: Jason <jasoli@nvidia.com> * bug fixes Signed-off-by: Jason <jasoli@nvidia.com> * v2 Signed-off-by: Jason <jasoli@nvidia.com> * changes Signed-off-by: Jason <jasoli@nvidia.com> * add types, old model working Signed-off-by: Jason <jasoli@nvidia.com> * pitch Signed-off-by: Jason <jasoli@nvidia.com> * update Signed-off-by: Jason <jasoli@nvidia.com> * update Signed-off-by: Jason <jasoli@nvidia.com> * let it work Signed-off-by: Jason <jasoli@nvidia.com> * fixes Signed-off-by: Jason <jasoli@nvidia.com> * add oktai comments Signed-off-by: Jason <jasoli@nvidia.com> * debug Signed-off-by: Jason <jasoli@nvidia.com> * scale Signed-off-by: Jason <jasoli@nvidia.com> * wip Signed-off-by: Jason <jasoli@nvidia.com> * fix test for v1 Signed-off-by: Jason <jasoli@nvidia.com> …
titu1994
added a commit
to titu1994/NeMo
that referenced
this pull request
Jul 20, 2021
* Itn add classes (#2141) * move do_training flag to config Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * added telephone to itn Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * add telephone and email to itn Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * ASR + NLP Doc Fixes (#2136) * Preserve the tokenizer config for ASR Signed-off-by: smajumdar <titu1994@gmail.com> * Correct nlp docs Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Removing graphsurgeon optional dependency, improving import error rep… (#2144) * Removing graphsurgeon optional dependency, improving import error reporting Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixing scope error Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix FilterbankFeatures eval nondeterminism. (#2146) Signed-off-by: PiotrDabkowski <piodrus@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix the docs. (#2148) Signed-off-by: Micha Livne <mlivne@nvidia.com> * Text processing refactor (#2149) * removed graphutils, suppletive, data_loader_utils from itn to be reused from tn Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * inheriting itn from tn, thus removing redundancy Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * cleaned whitelist Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * lgtm fix Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update how artifacts work (#2138) * Update how artifacts work Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * fixing some tests Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * fix more tests Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * add __init__ to tests to make them discoverable Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * empty src support Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * updates plust unittest Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * add copyright check Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * copyright header Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * fix style Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * handle hashed megatron checkpoint version in nlp restore_from Signed-off-by: ericharper <complex451@gmail.com> * add _MODEL_RESTORE_PATH to AppState Signed-off-by: ericharper <complex451@gmail.com> * get rid of global folder caching Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * double register - warning instead of exception Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * Add asr spe tests Signed-off-by: smajumdar <titu1994@gmail.com> * Pop out asr wpe pre-registered value Signed-off-by: smajumdar <titu1994@gmail.com> * Correct ASR tests and paths Signed-off-by: smajumdar <titu1994@gmail.com> * Correct tokenizer saving Signed-off-by: smajumdar <titu1994@gmail.com> * Correct ASR tests Signed-off-by: smajumdar <titu1994@gmail.com> * Correct ASR bpe mixin Signed-off-by: smajumdar <titu1994@gmail.com> * Patch up backward compatibility Signed-off-by: smajumdar <titu1994@gmail.com> * update register_bert_model Signed-off-by: ericharper <complex451@gmail.com> * update all get_lm_model calls Signed-off-by: ericharper <complex451@gmail.com> * return None if src not found Signed-off-by: ericharper <complex451@gmail.com> * handle case with no tokenizer Signed-off-by: ericharper <complex451@gmail.com> * do not add another hash is using tarfile_artifacts Signed-off-by: ericharper <complex451@gmail.com> * add return_none flag, update doc string Signed-off-by: ericharper <complex451@gmail.com> * update default behavior of register_artifact for NLPModel Signed-off-by: ericharper <complex451@gmail.com> * change kwarg name to verify_src_exists Signed-off-by: ericharper <complex451@gmail.com> * use cfg instead of _cfg Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * some cleanups Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Co-authored-by: ericharper <complex451@gmail.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Language model refactoring (#2120) * fixed branch in IR tutorial Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * bucketing tarred dataset for lm training Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * updated global rank Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * perplexity update Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * refactor lm to be campatible with latest nmt Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * perplexity change Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * removed obsolete config Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * added sequence perplexity Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * added non-smoothed CE loss for validation Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * unified sentence dataset, torchmetrics for sequence perplexity Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * translate_ddp refactor Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Signed-off-by: Micha Livne <mlivne@nvidia.com> * [NMT] Multi-validation Patch (#2150) * rename dl index 0 loss and sacrebleu for backwards compatibility Signed-off-by: ericharper <complex451@gmail.com> * eval -> val/tst Signed-off-by: ericharper <complex451@gmail.com> * instantiate torchmetrics after instantiating dataloaders Signed-off-by: ericharper <complex451@gmail.com> * bug Signed-off-by: ericharper <complex451@gmail.com> * remove debugging log Signed-off-by: ericharper <complex451@gmail.com> * remove debugging log Signed-off-by: ericharper <complex451@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * bumping version to 1.0.0 Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fixed the num_samples of text classification model. (#2152) Signed-off-by: Vahid <vnoroozi@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix for electronic (#2153) * fix for electronic Signed-off-by: ekmb <ebakhturina@nvidia.com> * special symbols added Signed-off-by: ekmb <ebakhturina@nvidia.com> * restrict symbols list Signed-off-by: ekmb <ebakhturina@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * FastSpeech 2 Test & Docs (#2143) * Add FS2 data loading test Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * TTS docs update for FastSpeech 2 Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Style fix for FS2 dataset test Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Fix transpose typo Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Minor patch for translate_ddp (#2155) * Patch for backtranslation in lm dataset Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * One more fix Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Entity linking (#2050) * Started adding SAP dataset Signed-off-by: Virginia Adams <vadams@nvidia.com> * Delete .lm_bert_dataset.py.swp Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added dataset and loss Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added entity linking encoder model Signed-off-by: Virginia Adams <vadams@nvidia.com> * Can build and use index from pubmedbert model Signed-off-by: Virginia Adams <vadams@nvidia.com> * checked boolean logic in build_index.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * End to end tested all functionality Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed val loss none at end of validation Signed-off-by: Virginia Adams <vadams@nvidia.com> * Started adding demo entity linking notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * adding in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * added call to entitylinking classes in __init__.py files Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added eval code to notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * Adding unfinished notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * Cleaned up example dir Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed recap commands Signed-off-by: Virginia Adams <vadams@nvidia.com> * added model typing and tiny data tar Signed-off-by: Virginia Adams <vadams@nvidia.com> * Adding tiny data zip Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated tiny example config data path Signed-off-by: Virginia Adams <vadams@nvidia.com> * Notebook demo works Signed-off-by: Virginia Adams <vadams@nvidia.com> * Changed training epochs Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed output from training and install cells Signed-off-by: Virginia Adams <vadams@nvidia.com> * changed code formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Started doc string for new functions Signed-off-by: Virginia Adams <vadams@nvidia.com> * Updated data_preprocessing to save to data_dir Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed comment in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * Update data_preprocessing.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated nemo typing imports Signed-off-by: Virginia Adams <vadams@nvidia.com> * about to rebase Signed-off-by: Virginia Adams <vadams@nvidia.com> * added back umls_dataset_processing.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed example data Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed typos in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed lgtm-com issues Signed-off-by: Virginia Adams <vadams@nvidia.com> * added copyright headers Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed import and copyright headers Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed formatting changes 2 Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed test formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added __init__.py for model and dataset Signed-off-by: Virginia Adams <vadams@nvidia.com> * loading newline file returns data_dir now Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed conf notebook and deleted comment Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * Updated Jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed file path Signed-off-by: Virginia Adams <vadams@nvidia.com> * Changed Jenkins pipeline order Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed Jenkins datapath... again... Signed-off-by: Virginia Adams <vadams@nvidia.com> * Made most review changes Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed copy right Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated unit test to wget config Signed-off-by: Virginia Adams <vadams@nvidia.com> * reverted test file back Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added project dir to jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * defined config in unit test Signed-off-by: Virginia Adams <vadams@nvidia.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Correct branch version for v1.0.0 (#2157) * Correct branch version Signed-off-by: smajumdar <titu1994@gmail.com> * Correct Jenkinsfile Signed-off-by: smajumdar <titu1994@gmail.com> * Update rst files Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * switch CI back to main Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fixed the docs. (#2156) Signed-off-by: Micha Livne <mlivne@nvidia.com> * Make Hifigan jittable (#2159) * FastSpeech 2 Test & Docs (#2143) * Add FS2 data loading test Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * TTS docs update for FastSpeech 2 Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Style fix for FS2 dataset test Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Fix transpose typo Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Jason <jasoli@nvidia.com> * Entity linking (#2050) * Started adding SAP dataset Signed-off-by: Virginia Adams <vadams@nvidia.com> * Delete .lm_bert_dataset.py.swp Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added dataset and loss Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added entity linking encoder model Signed-off-by: Virginia Adams <vadams@nvidia.com> * Can build and use index from pubmedbert model Signed-off-by: Virginia Adams <vadams@nvidia.com> * checked boolean logic in build_index.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * End to end tested all functionality Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed val loss none at end of validation Signed-off-by: Virginia Adams <vadams@nvidia.com> * Started adding demo entity linking notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * adding in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * added call to entitylinking classes in __init__.py files Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added eval code to notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * Adding unfinished notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * Cleaned up example dir Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed recap commands Signed-off-by: Virginia Adams <vadams@nvidia.com> * added model typing and tiny data tar Signed-off-by: Virginia Adams <vadams@nvidia.com> * Adding tiny data zip Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated tiny example config data path Signed-off-by: Virginia Adams <vadams@nvidia.com> * Notebook demo works Signed-off-by: Virginia Adams <vadams@nvidia.com> * Changed training epochs Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed output from training and install cells Signed-off-by: Virginia Adams <vadams@nvidia.com> * changed code formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Started doc string for new functions Signed-off-by: Virginia Adams <vadams@nvidia.com> * Updated data_preprocessing to save to data_dir Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed comment in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * Update data_preprocessing.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated nemo typing imports Signed-off-by: Virginia Adams <vadams@nvidia.com> * about to rebase Signed-off-by: Virginia Adams <vadams@nvidia.com> * added back umls_dataset_processing.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed example data Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed typos in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed lgtm-com issues Signed-off-by: Virginia Adams <vadams@nvidia.com> * added copyright headers Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed import and copyright headers Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed formatting changes 2 Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed test formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added __init__.py for model and dataset Signed-off-by: Virginia Adams <vadams@nvidia.com> * loading newline file returns data_dir now Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed conf notebook and deleted comment Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * Updated Jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed file path Signed-off-by: Virginia Adams <vadams@nvidia.com> * Changed Jenkins pipeline order Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed Jenkins datapath... again... Signed-off-by: Virginia Adams <vadams@nvidia.com> * Made most review changes Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed copy right Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated unit test to wget config Signed-off-by: Virginia Adams <vadams@nvidia.com> * reverted test file back Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added project dir to jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * defined config in unit test Signed-off-by: Virginia Adams <vadams@nvidia.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> * switch CI back to main Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * Make Hifigan jittable Signed-off-by: Ryan Leary <rleary@nvidia.com> * Remove vestigial debugging printout Signed-off-by: Ryan Leary <rleary@nvidia.com> * Add export forward and fix style Signed-off-by: Ryan Leary <rleary@nvidia.com> * Fix load_state_dict override for arbitrary layers Signed-off-by: Ryan Leary <rleary@nvidia.com> Co-authored-by: Jocelyn <jocelynh@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Jason <jasoli@nvidia.com> Co-authored-by: vadam5 <78445382+vadam5@users.noreply.github.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Co-authored-by: Ryan Leary <rleary@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix version (#2162) Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Megatron nb size reduced (#2163) * notebook size reduced Signed-off-by: ekmb <ebakhturina@nvidia.com> * notebook size reduced Signed-off-by: ekmb <ebakhturina@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update spectral clustering method (#2158) * update spectral clustering method Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * update Jenkins File Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * threshold fix by reducing window length for shorter embs Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * grammar fixes Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * CR update Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * paper reference Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * improve docstring for yaml Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * Doc fixes Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * revert (#2167) Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Limit Pytorch lightning release (#2170) Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * token classification models artifacts update (#2169) * artifacts update Signed-off-by: ekmb <ebakhturina@nvidia.com> * artifacts update Signed-off-by: ekmb <ebakhturina@nvidia.com> * jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> * fix for model restoration Signed-off-by: ekmb <ebakhturina@nvidia.com> * typos fix + jenkins dir update Signed-off-by: ekmb <ebakhturina@nvidia.com> * typo Signed-off-by: ericharper <complex451@gmail.com> * update jenkins branch Signed-off-by: ericharper <complex451@gmail.com> * add && Signed-off-by: ericharper <complex451@gmail.com> * jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> * jenkins disable Signed-off-by: ekmb <ebakhturina@nvidia.com> * revert jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> * jenkins disable Signed-off-by: ekmb <ebakhturina@nvidia.com> * revert jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> Co-authored-by: ericharper <complex451@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix to always_save_nemo (#2174) * Initial attempt at always_save_nemo fix Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * updated path before saving in exp manager, fixed bug when handling tarfile artifacts Signed-off-by: ericharper <complex451@gmail.com> * Add test with always_save_nemo to exp_manager Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * Style fixes Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * update jenkins branch Signed-off-by: ericharper <complex451@gmail.com> * check for nemo: Signed-off-by: ericharper <complex451@gmail.com> * check for nemo: Signed-off-by: ericharper <complex451@gmail.com> * check for nemo: Signed-off-by: ericharper <complex451@gmail.com> Co-authored-by: ericharper <complex451@gmail.com> Co-authored-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix typo (#2179) Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Make itn tests optional (#2173) * Limit Pytorch lightning release Signed-off-by: smajumdar <titu1994@gmail.com> * Add final two checks Signed-off-by: smajumdar <titu1994@gmail.com> Co-authored-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * First Revision of TTS Docs and Notebooks Update for 1.0 (#2166) * squash Signed-off-by: Jason <jasoli@nvidia.com> * notebook fixes Signed-off-by: Jason <jasoli@nvidia.com> * notebook fixes Signed-off-by: Jason <jasoli@nvidia.com> * typos Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * add more alternatives of 0 for telephone (#2171) Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Acc tn (#2180) * make tn cardinal faster Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * add number far Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * add test Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix lgtm Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix lgtm Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * [DOCS] NLP Model parallel, NMT multi-val, CORE register artifacts (#2168) * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Change label smoothing prob to reduce chance of test failure (#2184) Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Eric Harper <complex451@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add FS2 checkpoint links to docs and inference notebook (#2181) * Add FS2 checkpoint links to docs and inference notebook Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Remove empty cell from TTS notebook Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update ptl to 1.3 on main branch (#2178) * Update PTL Signed-off-by: smajumdar <titu1994@gmail.com> * Begin update to Pytorch Lightning 1.3.x Signed-off-by: smajumdar <titu1994@gmail.com> * Formatting Signed-off-by: smajumdar <titu1994@gmail.com> * style Signed-off-by: ericharper <complex451@gmail.com> * Formatting Signed-off-by: smajumdar <titu1994@gmail.com> * minor fix Signed-off-by: Jason <jasoli@nvidia.com> * minor fix Signed-off-by: Jason <jasoli@nvidia.com> * get testing attribute from trainer Signed-off-by: ericharper <complex451@gmail.com> * update init_ddp_connection override Signed-off-by: ericharper <complex451@gmail.com> * update attribute Signed-off-by: ericharper <complex451@gmail.com> * add barrier after load checkpoint in megatron Signed-off-by: ericharper <complex451@gmail.com> * remove barrier Signed-off-by: ericharper <complex451@gmail.com> * update last naming Signed-off-by: Jason <jasoli@nvidia.com> Co-authored-by: ericharper <complex451@gmail.com> Co-authored-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * SDE updates (#2187) * Added updates to SDE: - support for external vocabulary (to detect OOV words) - support for offset field (for segmented long recordings) - UI improvements Signed-off-by: Vitaly Lavrukhin <vlavrukhin@nvidia.com> * Refactored diff in SDE Signed-off-by: Vitaly Lavrukhin <vlavrukhin@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add TTS aligner and improved version of g2p for vocabs.Phonemes, small improvement in TalkNet (#2189) * add first version of aligner Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> * aligner docs, new g2p version, fix bugs in talknet Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> * update docs and remove lj related code Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> * fix style Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> * fix import Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * set the default of nodessplitter to None. (#2190) Signed-off-by: Vahid <vnoroozi@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * NMT fixes (#2194) * minor fixes Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * minor bugfixes Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Store mappings file in .nemo for FS2 model (#2196) * Store mappings file in .nemo for FS2 model Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Add error enforcing mappings file during training (FS2) Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add support to change the SE context window of ConvASREncoder (#2193) * Add support for changing context window on the fly Signed-off-by: smajumdar <titu1994@gmail.com> * Add support to change the SE context window of ConvASREncoder Signed-off-by: smajumdar <titu1994@gmail.com> * Add ability to skip config updating Signed-off-by: smajumdar <titu1994@gmail.com> * Switch to mixin based API Signed-off-by: smajumdar <titu1994@gmail.com> * Update docs and api for ASRModuleMixin Signed-off-by: smajumdar <titu1994@gmail.com> * Change print to logging.info Signed-off-by: smajumdar <titu1994@gmail.com> * Correct stride level when computing context window Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add a CI test for doing inference with an NMT model trained with Pre-LN (#2198) * Change label smoothing prob to reduce chance of test failure Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * Add Pre-LN inference test to Jenkinsfile Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * Separate tests for training and NMT inference Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix ipywidgets error in asr notebook (#2199) Added `ipywidgets` to avoid `ImportError: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html` error. Signed-off-by: Derek Chia <derek@derekchia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * metrics fix (#2202) * metrics fix Signed-off-by: ekmb <ebakhturina@nvidia.com> * metrics reset for punct model Signed-off-by: ekmb <ebakhturina@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * readme and minor improvements (#2203) * readme and minor improvements Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * vad threshold update Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix text processing docs (#2195) * fix text processing docs Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix name Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * add guard to pynini import Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix bug in SpecCutout (#2201) Signed-off-by: Robert Bracco <robertbracco1@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix bug in SpecCutout (#2201) (#2205) Signed-off-by: Robert Bracco <robertbracco1@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: smajumdar <titu1994@gmail.com> Co-authored-by: Robert Bracco <47190785+rbracco@users.noreply.github.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Set seed before generating random tensors in NMT test (#2206) * Change label smoothing prob to reduce chance of test failure Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * Set seed before generating tensors Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> Signed-off-by: Micha Livne <mlivne@nvidia.com> * ASR patches for v1.0.0 (#2207) * Multiple updates to RNNT add initialization Signed-off-by: smajumdar <titu1994@gmail.com> * Correct name of initilization Signed-off-by: smajumdar <titu1994@gmail.com> * Update dockerignore Signed-off-by: smajumdar <titu1994@gmail.com> * Fix RNNT WER calculation Signed-off-by: smajumdar <titu1994@gmail.com> * Address comments Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Multilingual training for NMT (#2160) * mnmt on fresh main Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * push for test Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * debug Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * check Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * cleanup Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * minor fix Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * more minor fixes Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * fix for test Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * fix list size error Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * multilingual in infer Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * changes Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * tar creation with multilingual Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * fix Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * changes + parallelism + bug fix Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * small fix Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * multilingual preprocessor fix Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * globally unique fragment names in tarred dataset Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * minor changes Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * rm load_from_cached_dataset Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * minor config change Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * rm unsued import Signed-off-by: Abhinav Khattar <aklife97@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Remove memory leak from ASR notebook + update model notebook (#2213) * ASR patches for v1.0.0 (#2207) * Multiple updates to RNNT add initialization Signed-off-by: smajumdar <titu1994@gmail.com> * Correct name of initilization Signed-off-by: smajumdar <titu1994@gmail.com> * Update dockerignore Signed-off-by: smajumdar <titu1994@gmail.com> * Fix RNNT WER calculation Signed-off-by: smajumdar <titu1994@gmail.com> * Address comments Signed-off-by: smajumdar <titu1994@gmail.com> * Correct model notebook to log the loss and correctly assign keys Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * replace names in vad tutorials (#2220) Signed-off-by: fayejf <fayejf07@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix the versioning name. (#2209) * fix the versioning name. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Made version None. Signed-off-by: Vahid <vnoroozi@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Enabled passing kwargs to export() (#2175) * Enabled passing kwargs to export() Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixing style; changed Classifier input_example to new extended syntax Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixed order of forward() call in export Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixing style Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update g2p: ambigious ignore, flag for skipping seq2seq (#2223) Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update TTS notebook with TalkNet inference (#2133) * Update TTS notebook with TalkNet inference. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Update TTS Notebook. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Update TTS TN Training Notebook. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Fix TN paper link. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Remove branch updaing TODOs. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update speaker notebooks (#2224) Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Support symlinked files (#2216) Signed-off-by: Anas Abou Allaban <aabouallaban@pm.me> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Set strict=True everywhere by default. (#2225) Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * set strict=True in nlp_model (#2227) Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * set strict=False for model parallel examples Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Make Text processing installation optional via reinstall.sh (#2226) * Make Text processing installation optional via reinstall.sh Signed-off-by: smajumdar <titu1994@gmail.com> * Support both success and failure states Signed-off-by: smajumdar <titu1994@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Transformer final norm preln (#2197) * fix pre_ln final norm Signed-off-by: fayejf <fayejf07@gmail.com> * style fix Signed-off-by: fayejf <fayejf07@gmail.com> * bug fixed Signed-off-by: fayejf <fayejf07@gmail.com> * bugfix post_ln Signed-off-by: fayejf <fayejf07@gmail.com> * update and add pre_ln_final_norm Signed-off-by: fayejf <fayejf07@gmail.com> * style fix Signed-off-by: fayejf <fayejf07@gmail.com> * fix for unit test Signed-off-by: fayejf <fayejf07@gmail.com> * rename final_norm to final_layer_norm Signed-off-by: fayejf <fayejf07@gmail.com> * bug fix Signed-off-by: fayejf <fayejf07@gmail.com> * tiny fix Signed-off-by: fayejf <fayejf07@gmail.com> * fix and improve Signed-off-by: fayejf <fayejf07@gmail.com> * tiny fix Signed-off-by: fayejf <fayejf07@gmail.com> * Patch for NMT to allow loading old modlels trained with pre-LN Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update models and notebook for 1.0 (#2211) * update models Signed-off-by: Jason <jasoli@nvidia.com> * updates Signed-off-by: Jason <jasoli@nvidia.com> * fix Signed-off-by: Jason <jasoli@nvidia.com> * add links Signed-off-by: Jason <jasoli@nvidia.com> * fixes Signed-off-by: Jason <jasoli@nvidia.com> * style Signed-off-by: Jason <jasoli@nvidia.com> * update checkpoints Signed-off-by: Jason <jasoli@nvidia.com> * rename Signed-off-by: Jason <jasoli@nvidia.com> * fixes Signed-off-by: Jason <jasoli@nvidia.com> * lgtm Signed-off-by: Jason <jasoli@nvidia.com> * fix loading waveglow Signed-off-by: Jason <jasoli@nvidia.com> * typo Signed-off-by: Jason <jasoli@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update_metrics_classification_models (#2228) Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Co-authored-by: fayejf <36722593+fayejf@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Data loader for seq of label model (#2084) * feature to seq label data loader Signed-off-by: fayejf <fayejf07@gmail.com> * style fix Signed-off-by: fayejf <fayejf07@gmail.com> * small fix Signed-off-by: fayejf <fayejf07@gmail.com> * update tl to be length of seq label Signed-off-by: fayejf <fayejf07@gmail.com> * style fix Signed-off-by: fayejf <fayejf07@gmail.com> * tiny bug fix Signed-off-by: fayejf <fayejf07@gmail.com> * small updates Signed-off-by: fayejf <fayejf07@gmail.com> * updates for review feedback Signed-off-by: fayejf <fayejf07@gmail.com> * style fix Signed-off-by: fayejf <fayejf07@gmail.com> * explain seq_label Signed-off-by: fayejf <fayejf07@gmail.com> * fix lgtm Signed-off-by: fayejf <fayejf07@gmail.com> * small updates Signed-off-by: fayejf <fayejf07@gmail.com> * improve as discussed Signed-off-by: fayejf <fayejf07@gmail.com> * add docstring Signed-off-by: fayejf <fayejf07@gmail.com> Co-authored-by: Nithin Rao <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix comments (#2236) * fix comments Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix typo Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * add paper ref to sgdqa model doc (#2233) * add paper ref to sgdqa model doc Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix comments Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Move ConcatDataset to common (#2237) * move concatdataset to common Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * var name change Signed-off-by: Abhinav Khattar <aklife97@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * audio based normalization (#2231) * squash norm_audio Signed-off-by: ekmb <ebakhturina@nvidia.com> * add missing files Signed-off-by: ekmb <ebakhturina@nvidia.com> * style Signed-off-by: ekmb <ebakhturina@nvidia.com> * unit tests added, docstrings fixed Signed-off-by: ekmb <ebakhturina@nvidia.com> * fix lgtm errors Signed-off-by: ekmb <ebakhturina@nvidia.com> * debug jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> * debug jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> * signature update Signed-off-by: ekmb <ebakhturina@nvidia.com> * set deterministic default Signed-off-by: ekmb <ebakhturina@nvidia.com> * add more test cases Signed-off-by: ekmb <ebakhturina@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * bug fix config (#2232) Signed-off-by: fayejf <fayejf07@gmail.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Alias Swish to SiLU (#2239) * Alias Swish to SiLU and move activations to inplace execution if possible Signed-off-by: smajumdar <titu1994@gmail.com> * Remove unused import Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update README.rst Signed-off-by: Micha Livne <mlivne@nvidia.com> * Offline asr notebook bug fix (#2242) * fix Signed-off-by: fayejf <fayejf07@gmail.com> * install Signed-off-by: fayejf <fayejf07@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix docstring (#2244) * fix comments Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix doc string Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update "last" Checkpoint (#2241) * fix Signed-off-by: Jason <jasoli@nvidia.com> * change Signed-off-by: Jason <jasoli@nvidia.com> * fix Signed-off-by: Jason <jasoli@nvidia.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add pretrained model stt_es_citrinet_512 (#2247) Signed-off-by: Elena Rastorgueva <erastorgueva@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * [BUGFIX] Only process tarfile artifacts when model was restored from tarfile (#2250) * process tarfile artifacts only if model is being restored Signed-off-by: ericharper <complex451@gmail.com> * process tarfile artifacts only if model was restored from a tarfile Signed-off-by: ericharper <complex451@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Log average metrics for Multi-validation in NMT (#2251) * add avg metrics NMT Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * name change Signed-off-by: Abhinav Khattar <aklife97@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update Primer notebook (#2258) Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fixed Bug 3310780 and 3310799 (#2264) Signed-off-by: Virginia Adams <vadams@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Support multiple models being instantiated in same execution scope (#2245) * Support multiple models being instantiated in same execution scope Signed-off-by: smajumdar <titu1994@gmail.com> * Fix tests Signed-off-by: smajumdar <titu1994@gmail.com> * Add locks to methods in appstate Signed-off-by: smajumdar <titu1994@gmail.com> * Perform locks only on write operations Signed-off-by: smajumdar <titu1994@gmail.com> * Correct deadlock issue Signed-off-by: smajumdar <titu1994@gmail.com> * Add more tests Signed-off-by: smajumdar <titu1994@gmail.com> * Add test for multi save and remove patch to change save type Signed-off-by: smajumdar <titu1994@gmail.com> * Update app state to preserve gidx of previous token Signed-off-by: smajumdar <titu1994@gmail.com> * Correct restoration logic for tarfiles Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * ASR Refactoring (#2240) * Refactor out the preprocessing from ASR into common Signed-off-by: smajumdar <titu1994@gmail.com> * Correct nltk issue with vocabs.py for clusters Signed-off-by: smajumdar <titu1994@gmail.com> * Add typing information to SpecAugment and SpecCutout Signed-off-by: smajumdar <titu1994@gmail.com> * Reorganize parts directory Signed-off-by: smajumdar <titu1994@gmail.com> * Refactor parts submodules, add __init__ to few important parts Signed-off-by: smajumdar <titu1994@gmail.com> * Update docs for new path to parts Signed-off-by: smajumdar <titu1994@gmail.com> * Cherry pick PR https://github.com/NVIDIA/NeMo/pull/2219 Signed-off-by: smajumdar <titu1994@gmail.com> * Add header for preprocessing commons Signed-off-by: smajumdar <titu1994@gmail.com> * Fix style of tests Signed-off-by: smajumdar <titu1994@gmail.com> * Add forced update of configs for train-val-test ds to new labels tests Signed-off-by: smajumdar <titu1994@gmail.com> * Update path to FilterbankFeatures for TTS Signed-off-by: smajumdar <titu1994@gmail.com> * Add an alias file for backward compatibility Signed-off-by: smajumdar <titu1994@gmail.com> * Add an alias file for backward compatibility Signed-off-by: smajumdar <titu1994@gmail.com> * Update training scripts of ASR to support finetuning Signed-off-by: smajumdar <titu1994@gmail.com> * Update Finetuning step to be ModelPT level Signed-off-by: smajumdar <titu1994@gmail.com> * Update docs for finetuning for ASR Signed-off-by: smajumdar <titu1994@gmail.com> * Fix style Signed-off-by: smajumdar <titu1994@gmail.com> * Update docs and scripts with fine-tuning info Signed-off-by: smajumdar <titu1994@gmail.com> * Update docs and scripts with fine-tuning info Signed-off-by: smajumdar <titu1994@gmail.com> * Fix style Signed-off-by: smajumdar <titu1994@gmail.com> * Update scripts Signed-off-by: smajumdar <titu1994@gmail.com> * Add comment for weight initialization Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * TTS Doc Fix and Remove TTS Test (#2272) * bug fix and remove test Signed-off-by: Jason <jasoli@nvidia.com> * syntax Signed-off-by: Jason <jasoli@nvidia.com> * syntax Signed-off-by: Jason <jasoli@nvidia.com> * syntax Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Talknet training Fix (#2273) * TalkNet Training notebook fix. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Remove debug stuff. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update (#2274) Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add links (#2275) * update Signed-off-by: Jason <jasoli@nvidia.com> * link Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Delete 3_TTS_TalkNet_Training.ipynb (#2276) Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * tune down logging (#2277) * tune down logging Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * debug message instead of removing it completely Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * minor bugfix Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * remove confusing message Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Restore TalkNet training notebook (#2281) * Restore TalkNet training notebook. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Remove torchaudio dep. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix ExpManager Issues and FastPitch (#2283) * backport exp_manager fixes to v1 Signed-off-by: Jason <jasoli@nvidia.com> * fix fastpitch Signed-off-by: Jason <jasoli@nvidia.com> * fix tests Signed-off-by: Jason <jasoli@nvidia.com> * update prefix Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Organize asr config folders (#2284) Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix and enable DALI tests (#2077) * Fix and enable DALI tests Signed-off-by: Joaquin Anton <janton@nvidia.com> * remove unused import Signed-off-by: Joaquin Anton <janton@nvidia.com> * Move DALI tests to a separate Jenkins stage Signed-off-by: Joaquin Anton <janton@nvidia.com> * Remove DALI tests from the main jenkins ASR stage Signed-off-by: Joaquin Anton <janton@nvidia.com> * Comment out MFCC test Signed-off-by: Joaquin Anton <janton@nvidia.com> * Working version Signed-off-by: Joaquin Anton <janton@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Added unit test for hifigan export, fixed hifigan export (#2279) * Added unit test for hifigan export, Removed runtime test from waveglow test (now in export) Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixed style Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixed style Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update conformer recipes (#2265) * updated readme asr. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added models. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed the docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed the docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed the docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * disabled test. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Updated the config files. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Updated the config files. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Updated the config files. Signed-off-by: Vahid <vnoroozi@nvidia.com> * dropped the wers. Signed-off-by: Vahid <vnoroozi@nvidia.com> * dropped the wers. Signed-off-by: Vahid <vnoroozi@nvidia.com> * dropped new models and reverted to old versions. Signed-off-by: Vahid <vnoroozi@nvidia.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Adding neural rescorer and its documentations (#2287) * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added more docs, figures, and output file. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added more docs, figures, and output file. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added more docs, figures, and output file. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added more docs, figures, and output file. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added more docs, figures, and output file. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed style Signed-off-by: Vahid <vnoroozi@nvidia.com> * add a note to asr notebook. Signed-off-by: Vahid <vnoroozi@nvidia.com> * add a note to asr notebook. Signed-off-by: Vahid <vnoroozi@nvidia.com> * add a note to asr notebook. Signed-off-by: Vahid <vnoroozi@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Adjust warning messages Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Revert "Adjust warning messages" This reverts commit df046ec55754d0136a2a28451435068f32409f30. Signed-off-by: Micha Livne <mlivne@nvidia.com> * Adjust warning messages (#2294) Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Adding new Models releases on NGC. (#2295) * added new models. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added tests for asr lm. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added tests for asr lm. Signed-off-by: Vahid <vnoroozi@nvidia.com> * dropped the test. Signed-off-by: Vahid <vnoroozi@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update quantization (#2298) Signed-off-by: slyned <slyned@nvidia.com> Co-authored-by: slyned <slyned@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * ASR improvements (#2293) * Update numba messages and citrinet configs Signed-off-by: smajumdar <titu1994@gmail.com> * Remove support for weight init scale and hidden hidden bias scale for layer normalized lstm Signed-off-by: smajumdar <titu1994@gmail.com> * Add support for multiple filetypes in tarred datasets, correct rnn LN-lstm inputs, fix OmegaConf compat issue Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Time quarter to (#2292) * fix comments Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix doc string Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * adding quarter to to time class Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix typo Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fixed paths. (#2301) Signed-off-by: Vahid <vnoroozi@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Added onnxruntime check of exported ONNX, bumped up default ONNX opset (#2278) * Added onnxruntime check of exported ONNX, bumped up default ONNX opset Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Made TS export to accept ONNX-style input example, removed unused param to export Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * check_trace default made False Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixed for updated export signature Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update readmes Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update readme Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update readme Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix docs table Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add support for Numba CUDA optimized SpecAugment (#2269) * Initial implementation Signed-off-by: smajumdar <titu1994@gmail.com> * Initial implementation Signed-off-by: smajumdar <titu1994@gmail.com> * Finish initial implementation of numba spec augment Signed-off-by: smajumdar <titu1994@gmail.com> * Correct mask propagataion Signed-off-by: smajumdar <titu1994@gmail.com> * Parallelize kernel over batch instead of over masks Signed-off-by: smajumdar <titu1994@gmail.com> * Finish tests and update to signature of spectrogramaugmentation calls Signed-off-by: smajumdar <titu1994@gmail.com> * Finish tests and update to signature of spectrogramaugmentation calls Signed-off-by: smajumdar <titu1994@gmail.com> * Add header Signed-off-by: smajumdar <titu1994@gmail.com> * Fix style Signed-off-by: smajumdar <titu1994@gmail.com> * Add heuristics Signed-off-by: smajumdar <titu1994@gmail.com> * Correct inclusive range of padding Signed-off-by: smajumdar <titu1994@gmail.com> * Correct typing for spec aug numba Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Added JSON manifest's support to transcribe_speech.py (#2304) * Added JSON manifest's support to transcribe_speech.py Signed-off-by: Vitaly Lavrukhin <vlavrukhin@nvidia.com> * Dropped unused import Signed-off-by: Vitaly Lavrukhin <vlavrukhin@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * get embedding for a single file (#2310) * get embedding for a single file Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * fixes Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * sr update Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * regain train mode Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update FastPitch (#2249) * wip Signed-off-by: Jason <jasoli@nvidia.com> * c1 Signed-off-by: Jason <jasoli@nvidia.com> * bug fixes Signed-off-by: Jason <jasoli@nvidia.com> * bug fixes Signed-off-by: Jason <jasoli@nvidia.com> * v2 Signed-off-by: Jason <jasoli@nvidia.com> * changes Signed-off-by: Jason <jasoli@nvidia.com> * add types, old model working Signed-off-by: Jason <jasoli@nvidia.com> * pitch Signed-off-by: Jason <jasoli@nvidia.com> * update Signed-off-by: Jason <jasoli@nvidia.com> * update Signed-off-by: Jason <jasoli@nvidia.com> * let it work Signed-off-by: Jason <jasoli@nvidia.com> * fixes Signed-off-by: Jason <jasoli@nvidia.com> * add oktai comments Signed-off-by: Jason <jasoli@nvidia.com> * debug Signed-off-by: Jason <jasoli@nvidia.com> * scale Signed-off-by: Jason <jasoli@nvidia.com> * wip Signed-off-by: Jason <jasoli@nvidia.com> * fix test for v1 Signed-off-by: Jason <jasoli@nvidia.com> * merge train and val Signed-off-by: Jason <jasoli@nvidia.com> * back to par bin att, add correct encoder settings Signed-off-by: Jason <jasoli@nvidia.com> * try Signed-off-by: Jason <jasoli@nvidia.com> * undo Signed-off-by: Jason <jasoli@nvidia.com> * lgtm: Signed-off-by: Jason <jasoli@nvidia.com> * style Signed-off-by: Jason <jasoli@nvidia.com> * default to ljs Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * patch quantization (#2314) * update quantization Signed-off-by: slyned <slyned@nvidia.com> * update quant infer trt Signed-off-by: slyned <slyned@nvidia.com> * fix style Signed-off-by: slyned <slyned@nvidia.com> Co-authored-by: slyned <slyned@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Pin OmegaConf version for 1.0.0 (#2316) * Update OmegaConf compatibility Signed-off-by: smajumdar <titu1994@gmail.com> * Correct OmegaConf.pretty() Signed-off-by: smajumdar <titu1994@gmail.com> * Upper bound omegaconf Signed-off-by: smajumdar <titu1994@gmail.com> * Revert "Correct OmegaConf.pretty()" This reverts commit 6ebae2ef Signed-off-by: smajumdar <titu1994@gmail.com> * Revert "Update OmegaConf compatibility" This reverts commit 83b2cf35a07a742552082e80e6ca34c9b8203cbc. Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * [BUGFIX] OmegaConf forward compatibility (#2319) * Update OmegaConf compatibility Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: ericharper <complex451@gmail.com> * Correct OmegaConf.pretty() Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: ericharper <complex451@gmail.com> * upper bound omegaconf Signed-off-by: ericharper <complex451@gmail.com> * add if,else back Signed-off-by: ericharper <complex451@gmail.com> * typo Signed-off-by: ericharper <complex451@gmail.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * bumping version to 1.0.1 Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix_cluster_small_sample (#2303) * fix_cluster_small_sample Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * for smaller samples Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * remove type Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * similarity matrix Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * est num of speakers add Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * comment update Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * style fix Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * MIN_SAMPLES passed through func arg Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * doc string update Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * spell mistake Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fastpitch export (#2300) * wip Signed-off-by: Jason <jasoli@nvidia.com> * c1 Signed-off-by: Jason <jasoli@nvidia.com> * bug fixes Signed-off-by: Jason <jasoli@nvidia.com> * bug fixes Signed-off-by: Jason <jasoli@nvidia.com> * v2 Signed-off-by: Jason <jasoli@nvidia.com> * changes Signed-off-by: Jason <jasoli@nvidia.com> * add types, old model working Signed-off-by: Jason <jasoli@nvidia.com> * pitch Signed-off-by: Jason <jasoli@nvidia.com> * update Signed-off-by: Jason <jasoli@nvidia.com> * update Signed-off-by: Jason <jasoli@nvidia.com> * let it work Signed-off-by: Jason <jasoli@nvidia.com> * fixes Signed-off-by: Jason <jasoli@nvidia.com> * add oktai comments Signed-off-by: Jason <jasoli@nvidia.com> * debug Signed-off-by: Jason <jasoli@nvidia.com> * scale Signed-off-by: Jason <jasoli@nvidia.com> * wip Signed-off-by: Jason <jasoli@nvidia.com> * fix test for v1 Signed-off-by: Jason <jasoli@nvidia.com> …
paarthneekhara
pushed a commit
to paarthneekhara/NeMo
that referenced
this pull request
Sep 17, 2021
* Itn add classes (#2141) * move do_training flag to config Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * added telephone to itn Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * add telephone and email to itn Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * ASR + NLP Doc Fixes (#2136) * Preserve the tokenizer config for ASR Signed-off-by: smajumdar <titu1994@gmail.com> * Correct nlp docs Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Removing graphsurgeon optional dependency, improving import error rep… (#2144) * Removing graphsurgeon optional dependency, improving import error reporting Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixing scope error Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix FilterbankFeatures eval nondeterminism. (#2146) Signed-off-by: PiotrDabkowski <piodrus@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix the docs. (#2148) Signed-off-by: Micha Livne <mlivne@nvidia.com> * Text processing refactor (#2149) * removed graphutils, suppletive, data_loader_utils from itn to be reused from tn Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * inheriting itn from tn, thus removing redundancy Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * cleaned whitelist Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * lgtm fix Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update how artifacts work (#2138) * Update how artifacts work Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * fixing some tests Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * fix more tests Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * add __init__ to tests to make them discoverable Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * empty src support Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * updates plust unittest Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * add copyright check Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * copyright header Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * fix style Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * handle hashed megatron checkpoint version in nlp restore_from Signed-off-by: ericharper <complex451@gmail.com> * add _MODEL_RESTORE_PATH to AppState Signed-off-by: ericharper <complex451@gmail.com> * get rid of global folder caching Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * double register - warning instead of exception Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * Add asr spe tests Signed-off-by: smajumdar <titu1994@gmail.com> * Pop out asr wpe pre-registered value Signed-off-by: smajumdar <titu1994@gmail.com> * Correct ASR tests and paths Signed-off-by: smajumdar <titu1994@gmail.com> * Correct tokenizer saving Signed-off-by: smajumdar <titu1994@gmail.com> * Correct ASR tests Signed-off-by: smajumdar <titu1994@gmail.com> * Correct ASR bpe mixin Signed-off-by: smajumdar <titu1994@gmail.com> * Patch up backward compatibility Signed-off-by: smajumdar <titu1994@gmail.com> * update register_bert_model Signed-off-by: ericharper <complex451@gmail.com> * update all get_lm_model calls Signed-off-by: ericharper <complex451@gmail.com> * return None if src not found Signed-off-by: ericharper <complex451@gmail.com> * handle case with no tokenizer Signed-off-by: ericharper <complex451@gmail.com> * do not add another hash is using tarfile_artifacts Signed-off-by: ericharper <complex451@gmail.com> * add return_none flag, update doc string Signed-off-by: ericharper <complex451@gmail.com> * update default behavior of register_artifact for NLPModel Signed-off-by: ericharper <complex451@gmail.com> * change kwarg name to verify_src_exists Signed-off-by: ericharper <complex451@gmail.com> * use cfg instead of _cfg Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * some cleanups Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Co-authored-by: ericharper <complex451@gmail.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Language model refactoring (#2120) * fixed branch in IR tutorial Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * bucketing tarred dataset for lm training Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * updated global rank Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * perplexity update Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * refactor lm to be campatible with latest nmt Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * perplexity change Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * removed obsolete config Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * added sequence perplexity Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * added non-smoothed CE loss for validation Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * unified sentence dataset, torchmetrics for sequence perplexity Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> * translate_ddp refactor Signed-off-by: AlexGrinch <grinchuk.alexey@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Signed-off-by: Micha Livne <mlivne@nvidia.com> * [NMT] Multi-validation Patch (#2150) * rename dl index 0 loss and sacrebleu for backwards compatibility Signed-off-by: ericharper <complex451@gmail.com> * eval -> val/tst Signed-off-by: ericharper <complex451@gmail.com> * instantiate torchmetrics after instantiating dataloaders Signed-off-by: ericharper <complex451@gmail.com> * bug Signed-off-by: ericharper <complex451@gmail.com> * remove debugging log Signed-off-by: ericharper <complex451@gmail.com> * remove debugging log Signed-off-by: ericharper <complex451@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * bumping version to 1.0.0 Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fixed the num_samples of text classification model. (#2152) Signed-off-by: Vahid <vnoroozi@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix for electronic (#2153) * fix for electronic Signed-off-by: ekmb <ebakhturina@nvidia.com> * special symbols added Signed-off-by: ekmb <ebakhturina@nvidia.com> * restrict symbols list Signed-off-by: ekmb <ebakhturina@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * FastSpeech 2 Test & Docs (#2143) * Add FS2 data loading test Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * TTS docs update for FastSpeech 2 Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Style fix for FS2 dataset test Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Fix transpose typo Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Minor patch for translate_ddp (#2155) * Patch for backtranslation in lm dataset Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * One more fix Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Entity linking (#2050) * Started adding SAP dataset Signed-off-by: Virginia Adams <vadams@nvidia.com> * Delete .lm_bert_dataset.py.swp Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added dataset and loss Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added entity linking encoder model Signed-off-by: Virginia Adams <vadams@nvidia.com> * Can build and use index from pubmedbert model Signed-off-by: Virginia Adams <vadams@nvidia.com> * checked boolean logic in build_index.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * End to end tested all functionality Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed val loss none at end of validation Signed-off-by: Virginia Adams <vadams@nvidia.com> * Started adding demo entity linking notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * adding in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * added call to entitylinking classes in __init__.py files Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added eval code to notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * Adding unfinished notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * Cleaned up example dir Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed recap commands Signed-off-by: Virginia Adams <vadams@nvidia.com> * added model typing and tiny data tar Signed-off-by: Virginia Adams <vadams@nvidia.com> * Adding tiny data zip Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated tiny example config data path Signed-off-by: Virginia Adams <vadams@nvidia.com> * Notebook demo works Signed-off-by: Virginia Adams <vadams@nvidia.com> * Changed training epochs Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed output from training and install cells Signed-off-by: Virginia Adams <vadams@nvidia.com> * changed code formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Started doc string for new functions Signed-off-by: Virginia Adams <vadams@nvidia.com> * Updated data_preprocessing to save to data_dir Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed comment in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * Update data_preprocessing.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated nemo typing imports Signed-off-by: Virginia Adams <vadams@nvidia.com> * about to rebase Signed-off-by: Virginia Adams <vadams@nvidia.com> * added back umls_dataset_processing.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed example data Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed typos in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed lgtm-com issues Signed-off-by: Virginia Adams <vadams@nvidia.com> * added copyright headers Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed import and copyright headers Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed formatting changes 2 Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed test formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added __init__.py for model and dataset Signed-off-by: Virginia Adams <vadams@nvidia.com> * loading newline file returns data_dir now Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed conf notebook and deleted comment Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * Updated Jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed file path Signed-off-by: Virginia Adams <vadams@nvidia.com> * Changed Jenkins pipeline order Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed Jenkins datapath... again... Signed-off-by: Virginia Adams <vadams@nvidia.com> * Made most review changes Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed copy right Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated unit test to wget config Signed-off-by: Virginia Adams <vadams@nvidia.com> * reverted test file back Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added project dir to jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * defined config in unit test Signed-off-by: Virginia Adams <vadams@nvidia.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Correct branch version for v1.0.0 (#2157) * Correct branch version Signed-off-by: smajumdar <titu1994@gmail.com> * Correct Jenkinsfile Signed-off-by: smajumdar <titu1994@gmail.com> * Update rst files Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * switch CI back to main Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fixed the docs. (#2156) Signed-off-by: Micha Livne <mlivne@nvidia.com> * Make Hifigan jittable (#2159) * FastSpeech 2 Test & Docs (#2143) * Add FS2 data loading test Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * TTS docs update for FastSpeech 2 Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Style fix for FS2 dataset test Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Fix transpose typo Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Jason <jasoli@nvidia.com> * Entity linking (#2050) * Started adding SAP dataset Signed-off-by: Virginia Adams <vadams@nvidia.com> * Delete .lm_bert_dataset.py.swp Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added dataset and loss Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added entity linking encoder model Signed-off-by: Virginia Adams <vadams@nvidia.com> * Can build and use index from pubmedbert model Signed-off-by: Virginia Adams <vadams@nvidia.com> * checked boolean logic in build_index.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * End to end tested all functionality Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed val loss none at end of validation Signed-off-by: Virginia Adams <vadams@nvidia.com> * Started adding demo entity linking notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * adding in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * added call to entitylinking classes in __init__.py files Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added eval code to notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * Adding unfinished notebook Signed-off-by: Virginia Adams <vadams@nvidia.com> * Cleaned up example dir Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed recap commands Signed-off-by: Virginia Adams <vadams@nvidia.com> * added model typing and tiny data tar Signed-off-by: Virginia Adams <vadams@nvidia.com> * Adding tiny data zip Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated tiny example config data path Signed-off-by: Virginia Adams <vadams@nvidia.com> * Notebook demo works Signed-off-by: Virginia Adams <vadams@nvidia.com> * Changed training epochs Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed output from training and install cells Signed-off-by: Virginia Adams <vadams@nvidia.com> * changed code formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Started doc string for new functions Signed-off-by: Virginia Adams <vadams@nvidia.com> * Updated data_preprocessing to save to data_dir Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed comment in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * Update data_preprocessing.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated nemo typing imports Signed-off-by: Virginia Adams <vadams@nvidia.com> * about to rebase Signed-off-by: Virginia Adams <vadams@nvidia.com> * added back umls_dataset_processing.py Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed example data Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed typos in notebook demo Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed lgtm-com issues Signed-off-by: Virginia Adams <vadams@nvidia.com> * added copyright headers Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed import and copyright headers Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed formatting changes 2 Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed test formatting Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added __init__.py for model and dataset Signed-off-by: Virginia Adams <vadams@nvidia.com> * loading newline file returns data_dir now Signed-off-by: Virginia Adams <vadams@nvidia.com> * Removed conf notebook and deleted comment Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * Updated Jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed file path Signed-off-by: Virginia Adams <vadams@nvidia.com> * Changed Jenkins pipeline order Signed-off-by: Virginia Adams <vadams@nvidia.com> * Fixed Jenkins datapath... again... Signed-off-by: Virginia Adams <vadams@nvidia.com> * Made most review changes Signed-off-by: Virginia Adams <vadams@nvidia.com> * fixed copy right Signed-off-by: Virginia Adams <vadams@nvidia.com> * updated unit test to wget config Signed-off-by: Virginia Adams <vadams@nvidia.com> * reverted test file back Signed-off-by: Virginia Adams <vadams@nvidia.com> * Added project dir to jenkins test Signed-off-by: Virginia Adams <vadams@nvidia.com> * defined config in unit test Signed-off-by: Virginia Adams <vadams@nvidia.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> * switch CI back to main Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * Make Hifigan jittable Signed-off-by: Ryan Leary <rleary@nvidia.com> * Remove vestigial debugging printout Signed-off-by: Ryan Leary <rleary@nvidia.com> * Add export forward and fix style Signed-off-by: Ryan Leary <rleary@nvidia.com> * Fix load_state_dict override for arbitrary layers Signed-off-by: Ryan Leary <rleary@nvidia.com> Co-authored-by: Jocelyn <jocelynh@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Jason <jasoli@nvidia.com> Co-authored-by: vadam5 <78445382+vadam5@users.noreply.github.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Co-authored-by: Ryan Leary <rleary@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix version (#2162) Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Megatron nb size reduced (#2163) * notebook size reduced Signed-off-by: ekmb <ebakhturina@nvidia.com> * notebook size reduced Signed-off-by: ekmb <ebakhturina@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update spectral clustering method (#2158) * update spectral clustering method Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * update Jenkins File Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * threshold fix by reducing window length for shorter embs Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * grammar fixes Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * CR update Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * paper reference Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * improve docstring for yaml Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * Doc fixes Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * revert (#2167) Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Limit Pytorch lightning release (#2170) Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * token classification models artifacts update (#2169) * artifacts update Signed-off-by: ekmb <ebakhturina@nvidia.com> * artifacts update Signed-off-by: ekmb <ebakhturina@nvidia.com> * jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> * fix for model restoration Signed-off-by: ekmb <ebakhturina@nvidia.com> * typos fix + jenkins dir update Signed-off-by: ekmb <ebakhturina@nvidia.com> * typo Signed-off-by: ericharper <complex451@gmail.com> * update jenkins branch Signed-off-by: ericharper <complex451@gmail.com> * add && Signed-off-by: ericharper <complex451@gmail.com> * jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> * jenkins disable Signed-off-by: ekmb <ebakhturina@nvidia.com> * revert jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> * jenkins disable Signed-off-by: ekmb <ebakhturina@nvidia.com> * revert jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> Co-authored-by: ericharper <complex451@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix to always_save_nemo (#2174) * Initial attempt at always_save_nemo fix Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * updated path before saving in exp manager, fixed bug when handling tarfile artifacts Signed-off-by: ericharper <complex451@gmail.com> * Add test with always_save_nemo to exp_manager Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * Style fixes Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * update jenkins branch Signed-off-by: ericharper <complex451@gmail.com> * check for nemo: Signed-off-by: ericharper <complex451@gmail.com> * check for nemo: Signed-off-by: ericharper <complex451@gmail.com> * check for nemo: Signed-off-by: ericharper <complex451@gmail.com> Co-authored-by: ericharper <complex451@gmail.com> Co-authored-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix typo (#2179) Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Make itn tests optional (#2173) * Limit Pytorch lightning release Signed-off-by: smajumdar <titu1994@gmail.com> * Add final two checks Signed-off-by: smajumdar <titu1994@gmail.com> Co-authored-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * First Revision of TTS Docs and Notebooks Update for 1.0 (#2166) * squash Signed-off-by: Jason <jasoli@nvidia.com> * notebook fixes Signed-off-by: Jason <jasoli@nvidia.com> * notebook fixes Signed-off-by: Jason <jasoli@nvidia.com> * typos Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * add more alternatives of 0 for telephone (#2171) Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Acc tn (#2180) * make tn cardinal faster Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * add number far Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * add test Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix lgtm Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix lgtm Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * [DOCS] NLP Model parallel, NMT multi-val, CORE register artifacts (#2168) * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> * update docs Signed-off-by: ericharper <complex451@gmail.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Change label smoothing prob to reduce chance of test failure (#2184) Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Eric Harper <complex451@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add FS2 checkpoint links to docs and inference notebook (#2181) * Add FS2 checkpoint links to docs and inference notebook Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Remove empty cell from TTS notebook Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update ptl to 1.3 on main branch (#2178) * Update PTL Signed-off-by: smajumdar <titu1994@gmail.com> * Begin update to Pytorch Lightning 1.3.x Signed-off-by: smajumdar <titu1994@gmail.com> * Formatting Signed-off-by: smajumdar <titu1994@gmail.com> * style Signed-off-by: ericharper <complex451@gmail.com> * Formatting Signed-off-by: smajumdar <titu1994@gmail.com> * minor fix Signed-off-by: Jason <jasoli@nvidia.com> * minor fix Signed-off-by: Jason <jasoli@nvidia.com> * get testing attribute from trainer Signed-off-by: ericharper <complex451@gmail.com> * update init_ddp_connection override Signed-off-by: ericharper <complex451@gmail.com> * update attribute Signed-off-by: ericharper <complex451@gmail.com> * add barrier after load checkpoint in megatron Signed-off-by: ericharper <complex451@gmail.com> * remove barrier Signed-off-by: ericharper <complex451@gmail.com> * update last naming Signed-off-by: Jason <jasoli@nvidia.com> Co-authored-by: ericharper <complex451@gmail.com> Co-authored-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * SDE updates (#2187) * Added updates to SDE: - support for external vocabulary (to detect OOV words) - support for offset field (for segmented long recordings) - UI improvements Signed-off-by: Vitaly Lavrukhin <vlavrukhin@nvidia.com> * Refactored diff in SDE Signed-off-by: Vitaly Lavrukhin <vlavrukhin@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add TTS aligner and improved version of g2p for vocabs.Phonemes, small improvement in TalkNet (#2189) * add first version of aligner Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> * aligner docs, new g2p version, fix bugs in talknet Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> * update docs and remove lj related code Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> * fix style Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> * fix import Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * set the default of nodessplitter to None. (#2190) Signed-off-by: Vahid <vnoroozi@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * NMT fixes (#2194) * minor fixes Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * minor bugfixes Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Store mappings file in .nemo for FS2 model (#2196) * Store mappings file in .nemo for FS2 model Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> * Add error enforcing mappings file during training (FS2) Signed-off-by: Jocelyn Huang <jocelynh@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add support to change the SE context window of ConvASREncoder (#2193) * Add support for changing context window on the fly Signed-off-by: smajumdar <titu1994@gmail.com> * Add support to change the SE context window of ConvASREncoder Signed-off-by: smajumdar <titu1994@gmail.com> * Add ability to skip config updating Signed-off-by: smajumdar <titu1994@gmail.com> * Switch to mixin based API Signed-off-by: smajumdar <titu1994@gmail.com> * Update docs and api for ASRModuleMixin Signed-off-by: smajumdar <titu1994@gmail.com> * Change print to logging.info Signed-off-by: smajumdar <titu1994@gmail.com> * Correct stride level when computing context window Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add a CI test for doing inference with an NMT model trained with Pre-LN (#2198) * Change label smoothing prob to reduce chance of test failure Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * Add Pre-LN inference test to Jenkinsfile Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * Separate tests for training and NMT inference Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix ipywidgets error in asr notebook (#2199) Added `ipywidgets` to avoid `ImportError: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html` error. Signed-off-by: Derek Chia <derek@derekchia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * metrics fix (#2202) * metrics fix Signed-off-by: ekmb <ebakhturina@nvidia.com> * metrics reset for punct model Signed-off-by: ekmb <ebakhturina@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * readme and minor improvements (#2203) * readme and minor improvements Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * vad threshold update Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix text processing docs (#2195) * fix text processing docs Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix name Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * add guard to pynini import Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix bug in SpecCutout (#2201) Signed-off-by: Robert Bracco <robertbracco1@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix bug in SpecCutout (#2201) (#2205) Signed-off-by: Robert Bracco <robertbracco1@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: smajumdar <titu1994@gmail.com> Co-authored-by: Robert Bracco <47190785+rbracco@users.noreply.github.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Set seed before generating random tensors in NMT test (#2206) * Change label smoothing prob to reduce chance of test failure Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> * Set seed before generating tensors Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> Signed-off-by: Micha Livne <mlivne@nvidia.com> * ASR patches for v1.0.0 (#2207) * Multiple updates to RNNT add initialization Signed-off-by: smajumdar <titu1994@gmail.com> * Correct name of initilization Signed-off-by: smajumdar <titu1994@gmail.com> * Update dockerignore Signed-off-by: smajumdar <titu1994@gmail.com> * Fix RNNT WER calculation Signed-off-by: smajumdar <titu1994@gmail.com> * Address comments Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Multilingual training for NMT (#2160) * mnmt on fresh main Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * push for test Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * debug Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * check Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * cleanup Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * minor fix Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * more minor fixes Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * fix for test Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * fix list size error Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * multilingual in infer Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * changes Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * tar creation with multilingual Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * fix Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * changes + parallelism + bug fix Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * small fix Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * multilingual preprocessor fix Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * globally unique fragment names in tarred dataset Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * minor changes Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * rm load_from_cached_dataset Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * minor config change Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * rm unsued import Signed-off-by: Abhinav Khattar <aklife97@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Remove memory leak from ASR notebook + update model notebook (#2213) * ASR patches for v1.0.0 (#2207) * Multiple updates to RNNT add initialization Signed-off-by: smajumdar <titu1994@gmail.com> * Correct name of initilization Signed-off-by: smajumdar <titu1994@gmail.com> * Update dockerignore Signed-off-by: smajumdar <titu1994@gmail.com> * Fix RNNT WER calculation Signed-off-by: smajumdar <titu1994@gmail.com> * Address comments Signed-off-by: smajumdar <titu1994@gmail.com> * Correct model notebook to log the loss and correctly assign keys Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * replace names in vad tutorials (#2220) Signed-off-by: fayejf <fayejf07@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix the versioning name. (#2209) * fix the versioning name. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Made version None. Signed-off-by: Vahid <vnoroozi@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Enabled passing kwargs to export() (#2175) * Enabled passing kwargs to export() Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixing style; changed Classifier input_example to new extended syntax Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixed order of forward() call in export Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixing style Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update g2p: ambigious ignore, flag for skipping seq2seq (#2223) Signed-off-by: Oktai Tatanov <oktai.tatanov@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update TTS notebook with TalkNet inference (#2133) * Update TTS notebook with TalkNet inference. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Update TTS Notebook. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Update TTS TN Training Notebook. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Fix TN paper link. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Remove branch updaing TODOs. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update speaker notebooks (#2224) Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Support symlinked files (#2216) Signed-off-by: Anas Abou Allaban <aabouallaban@pm.me> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Set strict=True everywhere by default. (#2225) Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * set strict=True in nlp_model (#2227) Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * set strict=False for model parallel examples Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Make Text processing installation optional via reinstall.sh (#2226) * Make Text processing installation optional via reinstall.sh Signed-off-by: smajumdar <titu1994@gmail.com> * Support both success and failure states Signed-off-by: smajumdar <titu1994@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Transformer final norm preln (#2197) * fix pre_ln final norm Signed-off-by: fayejf <fayejf07@gmail.com> * style fix Signed-off-by: fayejf <fayejf07@gmail.com> * bug fixed Signed-off-by: fayejf <fayejf07@gmail.com> * bugfix post_ln Signed-off-by: fayejf <fayejf07@gmail.com> * update and add pre_ln_final_norm Signed-off-by: fayejf <fayejf07@gmail.com> * style fix Signed-off-by: fayejf <fayejf07@gmail.com> * fix for unit test Signed-off-by: fayejf <fayejf07@gmail.com> * rename final_norm to final_layer_norm Signed-off-by: fayejf <fayejf07@gmail.com> * bug fix Signed-off-by: fayejf <fayejf07@gmail.com> * tiny fix Signed-off-by: fayejf <fayejf07@gmail.com> * fix and improve Signed-off-by: fayejf <fayejf07@gmail.com> * tiny fix Signed-off-by: fayejf <fayejf07@gmail.com> * Patch for NMT to allow loading old modlels trained with pre-LN Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update models and notebook for 1.0 (#2211) * update models Signed-off-by: Jason <jasoli@nvidia.com> * updates Signed-off-by: Jason <jasoli@nvidia.com> * fix Signed-off-by: Jason <jasoli@nvidia.com> * add links Signed-off-by: Jason <jasoli@nvidia.com> * fixes Signed-off-by: Jason <jasoli@nvidia.com> * style Signed-off-by: Jason <jasoli@nvidia.com> * update checkpoints Signed-off-by: Jason <jasoli@nvidia.com> * rename Signed-off-by: Jason <jasoli@nvidia.com> * fixes Signed-off-by: Jason <jasoli@nvidia.com> * lgtm Signed-off-by: Jason <jasoli@nvidia.com> * fix loading waveglow Signed-off-by: Jason <jasoli@nvidia.com> * typo Signed-off-by: Jason <jasoli@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update_metrics_classification_models (#2228) Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Co-authored-by: fayejf <36722593+fayejf@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Data loader for seq of label model (#2084) * feature to seq label data loader Signed-off-by: fayejf <fayejf07@gmail.com> * style fix Signed-off-by: fayejf <fayejf07@gmail.com> * small fix Signed-off-by: fayejf <fayejf07@gmail.com> * update tl to be length of seq label Signed-off-by: fayejf <fayejf07@gmail.com> * style fix Signed-off-by: fayejf <fayejf07@gmail.com> * tiny bug fix Signed-off-by: fayejf <fayejf07@gmail.com> * small updates Signed-off-by: fayejf <fayejf07@gmail.com> * updates for review feedback Signed-off-by: fayejf <fayejf07@gmail.com> * style fix Signed-off-by: fayejf <fayejf07@gmail.com> * explain seq_label Signed-off-by: fayejf <fayejf07@gmail.com> * fix lgtm Signed-off-by: fayejf <fayejf07@gmail.com> * small updates Signed-off-by: fayejf <fayejf07@gmail.com> * improve as discussed Signed-off-by: fayejf <fayejf07@gmail.com> * add docstring Signed-off-by: fayejf <fayejf07@gmail.com> Co-authored-by: Nithin Rao <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix comments (#2236) * fix comments Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix typo Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * add paper ref to sgdqa model doc (#2233) * add paper ref to sgdqa model doc Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix comments Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Move ConcatDataset to common (#2237) * move concatdataset to common Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * var name change Signed-off-by: Abhinav Khattar <aklife97@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * audio based normalization (#2231) * squash norm_audio Signed-off-by: ekmb <ebakhturina@nvidia.com> * add missing files Signed-off-by: ekmb <ebakhturina@nvidia.com> * style Signed-off-by: ekmb <ebakhturina@nvidia.com> * unit tests added, docstrings fixed Signed-off-by: ekmb <ebakhturina@nvidia.com> * fix lgtm errors Signed-off-by: ekmb <ebakhturina@nvidia.com> * debug jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> * debug jenkins Signed-off-by: ekmb <ebakhturina@nvidia.com> * signature update Signed-off-by: ekmb <ebakhturina@nvidia.com> * set deterministic default Signed-off-by: ekmb <ebakhturina@nvidia.com> * add more test cases Signed-off-by: ekmb <ebakhturina@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * bug fix config (#2232) Signed-off-by: fayejf <fayejf07@gmail.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Alias Swish to SiLU (#2239) * Alias Swish to SiLU and move activations to inplace execution if possible Signed-off-by: smajumdar <titu1994@gmail.com> * Remove unused import Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update README.rst Signed-off-by: Micha Livne <mlivne@nvidia.com> * Offline asr notebook bug fix (#2242) * fix Signed-off-by: fayejf <fayejf07@gmail.com> * install Signed-off-by: fayejf <fayejf07@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix docstring (#2244) * fix comments Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix doc string Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update "last" Checkpoint (#2241) * fix Signed-off-by: Jason <jasoli@nvidia.com> * change Signed-off-by: Jason <jasoli@nvidia.com> * fix Signed-off-by: Jason <jasoli@nvidia.com> Co-authored-by: Sandeep Subramanian <sandeep.subramanian.1@umontreal.ca> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add pretrained model stt_es_citrinet_512 (#2247) Signed-off-by: Elena Rastorgueva <erastorgueva@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * [BUGFIX] Only process tarfile artifacts when model was restored from tarfile (#2250) * process tarfile artifacts only if model is being restored Signed-off-by: ericharper <complex451@gmail.com> * process tarfile artifacts only if model was restored from a tarfile Signed-off-by: ericharper <complex451@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Log average metrics for Multi-validation in NMT (#2251) * add avg metrics NMT Signed-off-by: Abhinav Khattar <aklife97@gmail.com> * name change Signed-off-by: Abhinav Khattar <aklife97@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update Primer notebook (#2258) Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fixed Bug 3310780 and 3310799 (#2264) Signed-off-by: Virginia Adams <vadams@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Support multiple models being instantiated in same execution scope (#2245) * Support multiple models being instantiated in same execution scope Signed-off-by: smajumdar <titu1994@gmail.com> * Fix tests Signed-off-by: smajumdar <titu1994@gmail.com> * Add locks to methods in appstate Signed-off-by: smajumdar <titu1994@gmail.com> * Perform locks only on write operations Signed-off-by: smajumdar <titu1994@gmail.com> * Correct deadlock issue Signed-off-by: smajumdar <titu1994@gmail.com> * Add more tests Signed-off-by: smajumdar <titu1994@gmail.com> * Add test for multi save and remove patch to change save type Signed-off-by: smajumdar <titu1994@gmail.com> * Update app state to preserve gidx of previous token Signed-off-by: smajumdar <titu1994@gmail.com> * Correct restoration logic for tarfiles Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * ASR Refactoring (#2240) * Refactor out the preprocessing from ASR into common Signed-off-by: smajumdar <titu1994@gmail.com> * Correct nltk issue with vocabs.py for clusters Signed-off-by: smajumdar <titu1994@gmail.com> * Add typing information to SpecAugment and SpecCutout Signed-off-by: smajumdar <titu1994@gmail.com> * Reorganize parts directory Signed-off-by: smajumdar <titu1994@gmail.com> * Refactor parts submodules, add __init__ to few important parts Signed-off-by: smajumdar <titu1994@gmail.com> * Update docs for new path to parts Signed-off-by: smajumdar <titu1994@gmail.com> * Cherry pick PR https://github.com/NVIDIA/NeMo/pull/2219 Signed-off-by: smajumdar <titu1994@gmail.com> * Add header for preprocessing commons Signed-off-by: smajumdar <titu1994@gmail.com> * Fix style of tests Signed-off-by: smajumdar <titu1994@gmail.com> * Add forced update of configs for train-val-test ds to new labels tests Signed-off-by: smajumdar <titu1994@gmail.com> * Update path to FilterbankFeatures for TTS Signed-off-by: smajumdar <titu1994@gmail.com> * Add an alias file for backward compatibility Signed-off-by: smajumdar <titu1994@gmail.com> * Add an alias file for backward compatibility Signed-off-by: smajumdar <titu1994@gmail.com> * Update training scripts of ASR to support finetuning Signed-off-by: smajumdar <titu1994@gmail.com> * Update Finetuning step to be ModelPT level Signed-off-by: smajumdar <titu1994@gmail.com> * Update docs for finetuning for ASR Signed-off-by: smajumdar <titu1994@gmail.com> * Fix style Signed-off-by: smajumdar <titu1994@gmail.com> * Update docs and scripts with fine-tuning info Signed-off-by: smajumdar <titu1994@gmail.com> * Update docs and scripts with fine-tuning info Signed-off-by: smajumdar <titu1994@gmail.com> * Fix style Signed-off-by: smajumdar <titu1994@gmail.com> * Update scripts Signed-off-by: smajumdar <titu1994@gmail.com> * Add comment for weight initialization Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * TTS Doc Fix and Remove TTS Test (#2272) * bug fix and remove test Signed-off-by: Jason <jasoli@nvidia.com> * syntax Signed-off-by: Jason <jasoli@nvidia.com> * syntax Signed-off-by: Jason <jasoli@nvidia.com> * syntax Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Talknet training Fix (#2273) * TalkNet Training notebook fix. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Remove debug stuff. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update (#2274) Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add links (#2275) * update Signed-off-by: Jason <jasoli@nvidia.com> * link Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Delete 3_TTS_TalkNet_Training.ipynb (#2276) Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * tune down logging (#2277) * tune down logging Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> * debug message instead of removing it completely Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * minor bugfix Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * remove confusing message Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Restore TalkNet training notebook (#2281) * Restore TalkNet training notebook. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> * Remove torchaudio dep. Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix ExpManager Issues and FastPitch (#2283) * backport exp_manager fixes to v1 Signed-off-by: Jason <jasoli@nvidia.com> * fix fastpitch Signed-off-by: Jason <jasoli@nvidia.com> * fix tests Signed-off-by: Jason <jasoli@nvidia.com> * update prefix Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Organize asr config folders (#2284) Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fix and enable DALI tests (#2077) * Fix and enable DALI tests Signed-off-by: Joaquin Anton <janton@nvidia.com> * remove unused import Signed-off-by: Joaquin Anton <janton@nvidia.com> * Move DALI tests to a separate Jenkins stage Signed-off-by: Joaquin Anton <janton@nvidia.com> * Remove DALI tests from the main jenkins ASR stage Signed-off-by: Joaquin Anton <janton@nvidia.com> * Comment out MFCC test Signed-off-by: Joaquin Anton <janton@nvidia.com> * Working version Signed-off-by: Joaquin Anton <janton@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Added unit test for hifigan export, fixed hifigan export (#2279) * Added unit test for hifigan export, Removed runtime test from waveglow test (now in export) Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixed style Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixed style Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update conformer recipes (#2265) * updated readme asr. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added models. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed the docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed the docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed the docs. Signed-off-by: Vahid <vnoroozi@nvidia.com> * disabled test. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Updated the config files. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Updated the config files. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Updated the config files. Signed-off-by: Vahid <vnoroozi@nvidia.com> * dropped the wers. Signed-off-by: Vahid <vnoroozi@nvidia.com> * dropped the wers. Signed-off-by: Vahid <vnoroozi@nvidia.com> * dropped new models and reverted to old versions. Signed-off-by: Vahid <vnoroozi@nvidia.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Adding neural rescorer and its documentations (#2287) * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * Added intial neural rescorer scripts. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added more docs, figures, and output file. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added more docs, figures, and output file. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added more docs, figures, and output file. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added more docs, figures, and output file. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added more docs, figures, and output file. Signed-off-by: Vahid <vnoroozi@nvidia.com> * fixed style Signed-off-by: Vahid <vnoroozi@nvidia.com> * add a note to asr notebook. Signed-off-by: Vahid <vnoroozi@nvidia.com> * add a note to asr notebook. Signed-off-by: Vahid <vnoroozi@nvidia.com> * add a note to asr notebook. Signed-off-by: Vahid <vnoroozi@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Adjust warning messages Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Revert "Adjust warning messages" This reverts commit df046ec55754d0136a2a28451435068f32409f30. Signed-off-by: Micha Livne <mlivne@nvidia.com> * Adjust warning messages (#2294) Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Adding new Models releases on NGC. (#2295) * added new models. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added tests for asr lm. Signed-off-by: Vahid <vnoroozi@nvidia.com> * added tests for asr lm. Signed-off-by: Vahid <vnoroozi@nvidia.com> * dropped the test. Signed-off-by: Vahid <vnoroozi@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update quantization (#2298) Signed-off-by: slyned <slyned@nvidia.com> Co-authored-by: slyned <slyned@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * ASR improvements (#2293) * Update numba messages and citrinet configs Signed-off-by: smajumdar <titu1994@gmail.com> * Remove support for weight init scale and hidden hidden bias scale for layer normalized lstm Signed-off-by: smajumdar <titu1994@gmail.com> * Add support for multiple filetypes in tarred datasets, correct rnn LN-lstm inputs, fix OmegaConf compat issue Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Time quarter to (#2292) * fix comments Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix doc string Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * adding quarter to to time class Signed-off-by: Yang Zhang <yangzhang@nvidia.com> * fix typo Signed-off-by: Yang Zhang <yangzhang@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fixed paths. (#2301) Signed-off-by: Vahid <vnoroozi@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Added onnxruntime check of exported ONNX, bumped up default ONNX opset (#2278) * Added onnxruntime check of exported ONNX, bumped up default ONNX opset Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Made TS export to accept ONNX-style input example, removed unused param to export Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * check_trace default made False Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> * Fixed for updated export signature Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update readmes Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update readme Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * update readme Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix docs table Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Add support for Numba CUDA optimized SpecAugment (#2269) * Initial implementation Signed-off-by: smajumdar <titu1994@gmail.com> * Initial implementation Signed-off-by: smajumdar <titu1994@gmail.com> * Finish initial implementation of numba spec augment Signed-off-by: smajumdar <titu1994@gmail.com> * Correct mask propagataion Signed-off-by: smajumdar <titu1994@gmail.com> * Parallelize kernel over batch instead of over masks Signed-off-by: smajumdar <titu1994@gmail.com> * Finish tests and update to signature of spectrogramaugmentation calls Signed-off-by: smajumdar <titu1994@gmail.com> * Finish tests and update to signature of spectrogramaugmentation calls Signed-off-by: smajumdar <titu1994@gmail.com> * Add header Signed-off-by: smajumdar <titu1994@gmail.com> * Fix style Signed-off-by: smajumdar <titu1994@gmail.com> * Add heuristics Signed-off-by: smajumdar <titu1994@gmail.com> * Correct inclusive range of padding Signed-off-by: smajumdar <titu1994@gmail.com> * Correct typing for spec aug numba Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Added JSON manifest's support to transcribe_speech.py (#2304) * Added JSON manifest's support to transcribe_speech.py Signed-off-by: Vitaly Lavrukhin <vlavrukhin@nvidia.com> * Dropped unused import Signed-off-by: Vitaly Lavrukhin <vlavrukhin@nvidia.com> Co-authored-by: Oleksii Kuchaiev <okuchaiev@users.noreply.github.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * get embedding for a single file (#2310) * get embedding for a single file Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * fixes Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * sr update Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * regain train mode Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Update FastPitch (#2249) * wip Signed-off-by: Jason <jasoli@nvidia.com> * c1 Signed-off-by: Jason <jasoli@nvidia.com> * bug fixes Signed-off-by: Jason <jasoli@nvidia.com> * bug fixes Signed-off-by: Jason <jasoli@nvidia.com> * v2 Signed-off-by: Jason <jasoli@nvidia.com> * changes Signed-off-by: Jason <jasoli@nvidia.com> * add types, old model working Signed-off-by: Jason <jasoli@nvidia.com> * pitch Signed-off-by: Jason <jasoli@nvidia.com> * update Signed-off-by: Jason <jasoli@nvidia.com> * update Signed-off-by: Jason <jasoli@nvidia.com> * let it work Signed-off-by: Jason <jasoli@nvidia.com> * fixes Signed-off-by: Jason <jasoli@nvidia.com> * add oktai comments Signed-off-by: Jason <jasoli@nvidia.com> * debug Signed-off-by: Jason <jasoli@nvidia.com> * scale Signed-off-by: Jason <jasoli@nvidia.com> * wip Signed-off-by: Jason <jasoli@nvidia.com> * fix test for v1 Signed-off-by: Jason <jasoli@nvidia.com> * merge train and val Signed-off-by: Jason <jasoli@nvidia.com> * back to par bin att, add correct encoder settings Signed-off-by: Jason <jasoli@nvidia.com> * try Signed-off-by: Jason <jasoli@nvidia.com> * undo Signed-off-by: Jason <jasoli@nvidia.com> * lgtm: Signed-off-by: Jason <jasoli@nvidia.com> * style Signed-off-by: Jason <jasoli@nvidia.com> * default to ljs Signed-off-by: Jason <jasoli@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * patch quantization (#2314) * update quantization Signed-off-by: slyned <slyned@nvidia.com> * update quant infer trt Signed-off-by: slyned <slyned@nvidia.com> * fix style Signed-off-by: slyned <slyned@nvidia.com> Co-authored-by: slyned <slyned@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Pin OmegaConf version for 1.0.0 (#2316) * Update OmegaConf compatibility Signed-off-by: smajumdar <titu1994@gmail.com> * Correct OmegaConf.pretty() Signed-off-by: smajumdar <titu1994@gmail.com> * Upper bound omegaconf Signed-off-by: smajumdar <titu1994@gmail.com> * Revert "Correct OmegaConf.pretty()" This reverts commit 6ebae2ef Signed-off-by: smajumdar <titu1994@gmail.com> * Revert "Update OmegaConf compatibility" This reverts commit 83b2cf35a07a742552082e80e6ca34c9b8203cbc. Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * [BUGFIX] OmegaConf forward compatibility (#2319) * Update OmegaConf compatibility Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: ericharper <complex451@gmail.com> * Correct OmegaConf.pretty() Signed-off-by: smajumdar <titu1994@gmail.com> Signed-off-by: ericharper <complex451@gmail.com> * upper bound omegaconf Signed-off-by: ericharper <complex451@gmail.com> * add if,else back Signed-off-by: ericharper <complex451@gmail.com> * typo Signed-off-by: ericharper <complex451@gmail.com> Co-authored-by: Somshubra Majumdar <titu1994@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * bumping version to 1.0.1 Signed-off-by: Oleksii Kuchaiev <okuchaiev@nvidia.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * fix_cluster_small_sample (#2303) * fix_cluster_small_sample Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * for smaller samples Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * remove type Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * similarity matrix Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * est num of speakers add Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * comment update Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * style fix Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * MIN_SAMPLES passed through func arg Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * doc string update Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> * spell mistake Signed-off-by: nithinraok <nithinrao.koluguri@gmail.com> Signed-off-by: Micha Livne <mlivne@nvidia.com> * Fastpitch export (#2300) * wip Signed-off-by: Jason <jasoli@nvidia.com> * c1 Signed-off-by: Jason <jasoli@nvidia.com> * bug fixes Signed-off-by: Jason <jasoli@nvidia.com> * bug fixes Signed-off-by: Jason <jasoli@nvidia.com> * v2 Signed-off-by: Jason <jasoli@nvidia.com> * changes Signed-off-by: Jason <jasoli@nvidia.com> * add types, old model working Signed-off-by: Jason <jasoli@nvidia.com> * pitch Signed-off-by: Jason <jasoli@nvidia.com> * update Signed-off-by: Jason <jasoli@nvidia.com> * update Signed-off-by: Jason <jasoli@nvidia.com> * let it work Signed-off-by: Jason <jasoli@nvidia.com> * fixes Signed-off-by: Jason <jasoli@nvidia.com> * add oktai comments Signed-off-by: Jason <jasoli@nvidia.com> * debug Signed-off-by: Jason <jasoli@nvidia.com> * scale Signed-off-by: Jason <jasoli@nvidia.com> * wip Signed-off-by: Jason <jasoli@nvidia.com> * fix test for v1 Signed-off-by: Jason <jasoli@nvidia.com> …
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fix for config propagation in tutorial notebooks