Python utilities for deep learning research
python setup.py install
or if you prefer using pip
pip install .
It is recommended to use develop instead of install option to reflect changes in the directory
python setup.py develop
or if you prefer using pip
pip install -e .
python setup.py test
You can find researchutils documentation here
New features and bug fixes are welcome. Send PRs.
This project is using GitHub flow (See here for details) for development so do not try to push directly to master branch (It will be rejected anyway).
Target python versions are 2.7, 3.4, 3.5 and 3.6 (as of August 2018).
Use six, future or any other libraries to keep compatibility among above python versions.
Write your features under ./researchutils/
Write your tests for the features under ./tests/
Keep same directory structure of original chainer as much as possible under ./researchutils/chainer/.
For example, if you are writing new chainer.function, place your new function under
./researchutils/chainer/functions/xxx/
and write import statement in
./researchutils/chainer/functions/__init__.py
When adding new feature such as function/class, always and must write test(s) unless it will be rejected.
When writing tests, for example for feature_module.py, please create test module file of name test_feature_module_name.py and place exactly at the same layer of your feature module.
See below.
├── researchutils
│ ├── __init__.py
│ ...
│ ├── your_owesome_module.py
...
└── tests
├── __init__.py
...
├── test_your_owesome_module.py
│
...
Write documents of your new function/class/feature and explain what it does.
Writing documents is hard but helps others understanding what you implemented and how to use it.
We use numpy style docstring. When writing the docs, follow numpy style. See here for details.
Write your document in English.