Merge pull request #1 from NVIDIA/master

update repo

.gitignore

@@ -8,6 +8,8 @@ result
 tests/data/asr
 .DS_Store
 bert.pt.json
+work
+fastspeech_output
 
 # Byte-compiled / optimized / DLL files
 __pycache__/
@@ -21,6 +23,7 @@ __pycache__/
 # Distribution / packaging
 .idea
 .Python
+wandb
 build/
 develop-eggs/
 dist/
@@ -152,3 +155,9 @@ examples/*/wandb
 examples/*/data
 wandb
 dump.py
+
+docs/sources/source/test_build/
+
+# Checkpoints, config files and temporary files created in tutorials.
+examples/neural_graphs/*.chkpt
+examples/neural_graphs/*.yml

.readthedocs.yml

@@ -0,0 +1,31 @@
+# =============================================================================
+# Copyright (c) 2020 NVIDIA. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# =============================================================================
+
+# Read the Docs configuration file
+# See https://docs.readthedocs.io/en/stable/config-file/v2.html for details
+
+# Required field.
+version: 2
+
+# Build documentation in the docs/ directory with Sphinx.
+sphinx:
+  configuration: docs/sources/source/conf.py
+
+# Set the version of Python and requirements required to build your docs
+python:
+  version: 3.7
+  install:
+    - requirements: requirements/requirements_docs.txt

CHANGELOG.md

@@ -70,14 +70,69 @@ To release a new version, please update the changelog as followed:
 ## [Unreleased]
 
 ### Added
+- 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](https://github.com/NVIDIA/NeMo/pull/538)) - @yzhang123
+- Online Data Augmentation for ASR Collection. ([PR #565](https://github.com/NVIDIA/NeMo/pull/565)) - @titu1994
+- Speed augmentation on CPU, TimeStretch augmentation on CPU+GPU ([PR #594](https://github.com/NVIDIA/NeMo/pull/565)) - @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](https://github.com/NVIDIA/NeMo/pull/602))
+- Online audio augmentation notebook in ASR examples ([PR #605](https://github.com/NVIDIA/NeMo/pull/605)) - @titu1994
+- ContextNet Encoder + Decoder Initial Support ([PR #630](https://github.com/NVIDIA/NeMo/pull/630)) - @titu1994
+- Added finetuning with Megatron-LM ([PR #601](https://github.com/NVIDIA/NeMo/pull/601)) - @ekmb
+- Added documentation for 8 kHz model ([PR #632](https://github.com/NVIDIA/NeMo/pull/632)) - @jbalam-nv
+
+
+### Changed
+- 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](https://github.com/NVIDIA/NeMo/pull/637)) - @redoctopus
+- Updated the callback system. Old callbacks will be deprecated in version 0.12. ([PR #615](https://github.com/NVIDIA/NeMo/pull/615)) - @blisc
+
+### Dependencies Update
+
+### Deprecated
+
+### Fixed
+
+### Removed
+
+### Security
+
+### Contributors
+
+## [0.10.2] - 2020-05-05
+
+### Added
+- 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](https://github.com/NVIDIA/NeMo/pull/413)) - @tkornuta
+
+## [0.10.0] - 2020-04-03
+
+### Added
+- Roberta and Albert support added to GLUE script, data caching also added.
+([PR #413](https://github.com/NVIDIA/NeMo/pull/413)) - @ekmb
+- text classification notebook added
+([PR #382](https://github.com/NVIDIA/NeMo/pull/382)) - @ericharper
+- New Neural Type System documentation. Also added decorator to generate docs for input/output ports.
+([PR #370](https://github.com/NVIDIA/NeMo/pull/370)) - @okuchaiev
 - New Neural Type System and its tests.
 ([PR #307](https://github.com/NVIDIA/NeMo/pull/307)) - @okuchaiev
 - Named tensors tuple module's output for graph construction.
 ([PR #268](https://github.com/NVIDIA/NeMo/pull/268)) - @stasbel
 - Introduced the `deprecated` decorator.
 ([PR #298](https://github.com/NVIDIA/NeMo/pull/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](https://github.com/NVIDIA/NeMo/pull/339)) - @tkornuta-nvidia
+- Speech Commands support.
+([PR #375](https://github.com/NVIDIA/NeMo/pull/375)) - @titu1994
 
 ### Changed
+- Refactoring of `nemo_nlp` collections:
+([PR #368](https://github.com/NVIDIA/NeMo/pull/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](https://github.com/NVIDIA/NeMo/pull/316)) - @VahidooX, @yzhang123, @ekmb
+    - renaming and restructuring of files, folder, and functions in `nemo_nlp`
+    - Updated licenses
 - All collections changed to use New Neural Type System.
 ([PR #307](https://github.com/NVIDIA/NeMo/pull/307)) - @okuchaiev
 - Additional Collections Repositories merged into core `nemo_toolkit` package.
@@ -86,28 +141,30 @@ To release a new version, please update the changelog as followed:
 ([PR #284](https://github.com/NVIDIA/NeMo/pull/284)) - @stasbel
 - NeMo is not longer using pep8 code style rules. Code style rules are now enforced with `isort` and `black` incorporated into CI checks.
 ([PR #286](https://github.com/NVIDIA/NeMo/pull/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).  
+- 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](https://github.com/NVIDIA/NeMo/pull/309)) - @tkornuta-nvidia
-- Refactoring of `nemo_nlp` collections: 
-([PR #316](https://github.com/NVIDIA/NeMo/pull/316)) - @VahidooX, @yzhang123, @ekmb
-    - renaming of files and restructuring of folder in `nemo_nlp`
-    - Updated licenses
 - 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](https://github.com/NVIDIA/NeMo/pull/267), [PR #283](https://github.com/NVIDIA/NeMo/pull/283), [PR #305](https://github.com/NVIDIA/NeMo/pull/305), [PR #311](https://github.com/NVIDIA/NeMo/pull/311)) - @blisc
 - Changed Distributed Data Parallel from Apex to Torch
 ([PR #336](https://github.com/NVIDIA/NeMo/pull/336)) - @blisc
 
 - Added TRADE (dialogue state tracking model) on MultiWOZ dataset
 ([PR #322](https://github.com/NVIDIA/NeMo/pull/322)) - @chiphuyen, @VahidooX
+- Question answering:
+([PR #390](https://github.com/NVIDIA/NeMo/pull/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
 
 ### Dependencies Update
 - Added dependency on `wrapt` (the new version of the `deprecated` warning) - @tkornuta-nvidia, @DEKHTIARJonathan
 
 ### Deprecated
 
 ### Fixed
-- Critical fix of the training action on CPU 
+- Critical fix of the training action on CPU
 ([PR #308](https://github.com/NVIDIA/NeMo/pull/309)) - @tkornuta-nvidia
+- Fixed issue in Tacotron 2 prenet
+([PR #444](https://github.com/NVIDIA/NeMo/pull/444)) - @blisc
 
 ### Removed
 - gradient_predivide_factor arg of train() now has no effect
@@ -166,7 +223,8 @@ This release also includes nemo_asr'' and nemo_nlp'' collections for Speech Reco
 
 Please refer to the documentation here: https://nvidia.github.io/NeMo/
 
-[Unreleased]: https://github.com/NVIDIA/NeMo/compare/v0.9.0...master
+[Unreleased]: https://github.com/NVIDIA/NeMo/compare/v0.10.0...master
+[0.10.0]: https://github.com/NVIDIA/NeMo/compare/v0.9.0...v0.10.0
 [0.9.0]: https://github.com/NVIDIA/NeMo/compare/v0.8.2...v0.9.0
 [0.8.2]: https://github.com/NVIDIA/NeMo/compare/v0.8.1...v0.8.2
 [0.8.1]: https://github.com/NVIDIA/NeMo/compare/r0.8...v0.8.1

CONTRIBUTING.md

@@ -4,9 +4,9 @@
 
 2) Make sure you sign your commits. E.g. use ``git commit -s`` when commiting
 
-3) Make sure all unittests finish successfully before sending PR
+3) Make sure all unittests finish successfully before sending PR ``python -m unittest`` from NeMo's root folder
 
-4) Send your Pull Request to `master` branch
+4) Send your Pull Request to the `master` branch
 
 
 # Collection Guidelines
@@ -28,9 +28,8 @@ Please note that CI needs to pass for all the modules and collections.
 1. **Sensible**: code should make sense. If you think a piece of code might be confusing, write comments.
 
 ## Python style
-We follow [PEP 8 style guide](https://www.python.org/dev/peps/pep-0008/) and we incorporate [pycodestyle](https://pypi.org/project/pycodestyle/) into our CI pipeline to check for style. Make sure that your code passes PEP 8 before creating a Pull Request.
-
-There are several tools to automatically format your code to be PEP 8 compliant, such as [autopep8](https://github.com/hhatto/autopep8). Your text editor might support its own auto PEP 8 plugin.
+We use ``black`` as our style guide. To check whether your code will pass style check (from the NeMo's repo folder) run:
+``python setup.py style`` and if it does not pass run ``python setup.py style --fix``.
 
 1. Avoid wild import: ``from X import *`` unless in ``X.py``, ``__all__`` is defined.
 1. Minimize the use of ``**kwargs``.
@@ -47,7 +46,10 @@ There are several tools to automatically format your code to be PEP 8 compliant,
 1. If a comment lasts multiple lines, use ``'''`` instead of ``#``.
 
 ## Nemo style
-1. If you import a module from the same collection, use relative path instead of absolute path. For example, inside ``nemo_nlp``, use ``.utils`` instead of ``nemo_nelp.utils``.
+1. Use absolute paths.
 1. Before accessing something, always make sure that it exists.
 1. Right inheritance. For example, if a module doesn't have any trainable weights, don't inherit from TrainableNM.
 1. Naming consistency, both within NeMo and between NeMo and external literature. E.g. use the name ``logits`` for ``log_probs``, ``hidden_size`` for ``d_model``.
+1. Make an effort to use the right Neural Types when designing your neural modules. If a type you need does not
+ exists - you can introduce one. See documentation on how to do this
+1. When creating input/ouput ports for your modules use "add_port_docs" decorator to nicely generate docs for them

Dockerfile

@@ -30,14 +30,23 @@ RUN apt-get update && \
     python-dev && \
     rm -rf /var/lib/apt/lists/*
 
-# install onnx trt open source plugins
+# install trt
 ENV PATH=$PATH:/usr/src/tensorrt/bin
-WORKDIR /tmp/onnx-trt
-COPY scripts/docker/onnx-trt.patch .
-RUN git clone -n https://github.com/onnx/onnx-tensorrt.git && cd onnx-tensorrt && \
-    git checkout 8716c9b && git submodule update --init --recursive && patch -f < ../onnx-trt.patch && \
-    mkdir build && cd build && cmake .. -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=/usr -DGPU_ARCHS="60 70 75" && \
-    make -j16 && make install && mv -f /usr/lib/libnvonnx* /usr/lib/x86_64-linux-gnu/ && ldconfig && rm -rf /tmp/onnx-tensorrt
+WORKDIR /tmp/trt-oss
+ARG NV_REPO=https://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1804/x86_64
+
+RUN cd /tmp/trt-oss
+ARG DEB=libcudnn7_7.6.5.32-1+cuda10.2_amd64.deb
+RUN curl -sL --output ${DEB} ${NV_REPO}/${DEB}
+ARG DEB=libnvinfer7_7.0.0-1+cuda10.2_amd64.deb
+RUN curl -sL --output ${DEB} ${NV_REPO}/${DEB}
+ARG DEB=libnvinfer-plugin7_7.0.0-1+cuda10.2_amd64.deb
+RUN curl -sL --output ${DEB} ${NV_REPO}/${DEB}
+ARG DEB=libnvonnxparsers7_7.0.0-1+cuda10.2_amd64.deb
+RUN curl -sL --output ${DEB} ${NV_REPO}/${DEB}
+ARG DEB=python-libnvinfer_7.0.0-1+cuda10.2_amd64.deb
+RUN curl -sL --output ${DEB} ${NV_REPO}/${DEB}
+RUN dpkg -i *.deb && cd ../.. && rm -rf /tmp/trt-oss
 
 # install nemo dependencies
 WORKDIR /tmp/nemo
@@ -52,6 +61,7 @@ COPY . .
 FROM nemo-deps as nemo
 ARG NEMO_VERSION
 ARG BASE_IMAGE
+
 # Check that NEMO_VERSION is set. Build will fail without this. Expose NEMO and base container
 # version information as runtime environment variable for introspection purposes
 RUN /usr/bin/test -n "$NEMO_VERSION" && \