All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog and this project adheres to Semantic Versioning.
- Added NeMoModels class. Implemented in ASR collection: ASRConvCTCModel, and QuartzNet and JasperNet as its children - @okuchaiev
- Added multi-dataset data-layer and dataset. (PR #538) - @yzhang123
- Online Data Augmentation for ASR Collection. (PR #565) - @titu1994
- Speed augmentation on CPU, TimeStretch augmentation on CPU+GPU (PR #594) - @titu1994
- Added TarredAudioToTextDataLayer, which allows for loading ASR datasets with tarred audio. Existing datasets can be converted with the
convert_to_tarred_audio_dataset.py
script. (PR #602) - Online audio augmentation notebook in ASR examples (PR #605) - @titu1994
- ContextNet Encoder + Decoder Initial Support (PR #630) - @titu1994
- Added finetuning with Megatron-LM (PR #601) - @ekmb
- Added documentation for 8 kHz model (PR #632) - @jbalam-nv
- The Neural Graph is a high-level abstract concept empowering the users to build graphs consisting of many, interconnected Neural Modules. A user in his/her application can build any number of graphs, potentially spanning over the same modules. The import/export options combined with the lightweight API make Neural Graphs a perfect tool for rapid prototyping and experimentation. (PR #413) - @tkornuta-nvidia
- Created the NeMo CV collection, added MNIST and CIFAR10 thin datalayers, implemented/ported several general usage trainable and non-trainable modules, added several new ElementTypes (PR #654) - @tkornuta-nvidia
- Added SGD dataset and SGD model baseline (PR #612) - @ekmb
- Policy Manager and Natural Language Generation Modules for MultiWOZ added (PR #691) - @ekmb
- quartznet and jasper ASR examples reworked into speech2text.py and speech2text_infer.py - @okuchaiev
- Syncs across workers at each step to check for NaN or inf loss. Terminates all workers if stop_on_nan_loss is set (as before), lets Apex deal with it if apex.amp optimization level is O1 or higher, and skips the step across workers otherwise. (PR #637) - @redoctopus
- Updated the callback system. Old callbacks will be deprecated in version 0.12. (PR #615) - @blisc
0.10.0 - 2020-04-03
- Roberta and Albert support added to GLUE script, data caching also added. (PR #413) - @ekmb
- text classification notebook added (PR #382) - @ericharper
- New Neural Type System documentation. Also added decorator to generate docs for input/output ports. (PR #370) - @okuchaiev
- New Neural Type System and its tests. (PR #307) - @okuchaiev
- Named tensors tuple module's output for graph construction. (PR #268) - @stasbel
- Introduced the
deprecated
decorator. (PR #298) - @tkornuta-nvidia - Implemented new mechanisms for importing and exporting of module configuration (init_params) to configuration (yml) files, along with unit tests, examples and tutorials (PR #339) - @tkornuta-nvidia
- Speech Commands support. (PR #375) - @titu1994
-
Refactoring of
nemo_nlp
collections: (PR #368) - @VahidooX, @yzhang123, @ekmb- renaming and restructuring of files, folder, and functions in
nemo_nlp
- losses cleaned up. LossAggregatorNM moved to nemo/backends/pytorch/common/losses (PR #316) - @VahidooX, @yzhang123, @ekmb
- renaming and restructuring of files, folder, and functions in
nemo_nlp
- Updated licenses
- renaming and restructuring of files, folder, and functions in
-
All collections changed to use New Neural Type System. (PR #307) - @okuchaiev
-
Additional Collections Repositories merged into core
nemo_toolkit
package. (PR #289) - @DEKHTIARJonathan -
Refactor manifest files parsing and processing for re-using. (PR #284) - @stasbel
-
NeMo is not longer using pep8 code style rules. Code style rules are now enforced with
isort
andblack
incorporated into CI checks. (PR #286) - @stasbel -
Major cleanup of Neural Module constructors (init), aiming at increasing the framework robustness: cleanup of NeuralModule initialization logic, refactor of trainer/actions (getting rid of local_params), fixes of several examples and unit tests, extraction and storing of intial parameters (init_params). (PR #309) - @tkornuta-nvidia
-
Updated nemo's use of the logging library. from nemo import logging is now the reccomended way of using the nemo logger. neural_factory.logger and all other instances of logger are now deprecated and planned for removal in the next version. Please see PR 267 for complete change information. (PR #267, PR #283, PR #305, PR #311) - @blisc
-
Changed Distributed Data Parallel from Apex to Torch (PR #336) - @blisc
-
Added TRADE (dialogue state tracking model) on MultiWOZ dataset (PR #322) - @chiphuyen, @VahidooX
-
Question answering: (PR #390) - @yzhang123
- Changed question answering task to use Roberta and Albert as alternative backends to Bert
- Added inference mode that does not require ground truth labels
- Added dependency on
wrapt
(the new version of thedeprecated
warning) - @tkornuta-nvidia, @DEKHTIARJonathan
- Critical fix of the training action on CPU (PR #308) - @tkornuta-nvidia
- Fixed issue in Tacotron 2 prenet (PR #444) - @blisc
- gradient_predivide_factor arg of train() now has no effect (PR #336) - @blisc
- Dropped support of the following ASR configs: jasper10x4.yaml, quartznet10x5.yaml, quartznet15x5_in.yaml, quartznet5x3.yaml, quartznet5x5.yaml, quartznet_an4.yaml. They are moved to experimental/configs and can still be used with v0.9 for use in replicating paper results (PR #354) - @blisc
0.9.0 - 2019-12-16
This release contains new features, new models and quality improvements for NeMo.
- Added "nemo_tts" - a Speech Synthesis collection with necessary modules for Tacotron2 and WaveGlow
- Added Mandarin support into nemo_asr and nemo_nlp
- Updated ASR and TTS checkpoints including Mandarin ASR
- Documentation now translated to Mandarin https://nvidia.github.io/NeMo/chinese/intro.html
- Export functionality for deployment
- General improvements and bug-fixes
0.8.2 - 2019-11-14
This is a quality improvement release for NeMo.
- Bugfixes
- Support for Pytorch 1.3
0.8.1 - 2019-12-16
This is a quality improvement release for NeMo.
- Added introductory ASR tutorial explaining how to get started with deep learning for ASR
- Re-organization of NeMo NLP library
- More efficient BERT pre-training implementation
- General improvements and bugfixes
- Support for CPU-only scenario
- David Pollack @dhpollack
- Harisankar Haridas @harisankarh
- Dilshod Tadjibaev @antimora
0.8.0 - 2019-12-16
The first public release of NVIDIA Neural Modules: NeMo.
This release also includes nemo_asr'' and nemo_nlp'' collections for Speech Recognition and Natural Language Processing.
Please refer to the documentation here: https://nvidia.github.io/NeMo/