The fastai deep learning library.
Copyright 2017 onwards, fast.ai, Inc. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. A copy of the License is provided in the LICENSE file in this repository.
This is a ground-up rewrite of fastai. Everything should work, although docs are still in progress. If you're interested in contributing, join the discussion at: http://forums.fast.ai/c/fastai-dev.
To use the notebooks or the beta version of the fastai modules you will need:
- to use python 3.7 or python 3.6 with dataclasses:
pip install dataclasses - to use the
pytorch-nightlyconda package, or the master branch of pytorch master - to install fastprogress:
pip install fastprogress
First install the nightly pytorch build, e.g. for CUDA 9.2:
pip install torch_nightly -f https://download.pytorch.org/whl/nightly/cu92/torch_nightly.html
If you have a different CUDA version find the right build here. Choose Preview/Linux/Pip/python3.6 and Your CUDA version and it will give you the correct install instruction.
Now you can install fastai. Note, that this is a beta test version at the moment, please report any issues:
pip install --index-url https://test.pypi.org/simple/ --extra-index-url https://pypi.org/simple/ fastai==1.0.0b3
First install the nightly pytorch build:
conda install -c pytorch pytorch-nightly
Now you can install fastai. Note, that this is a beta test version at the moment, please report any issues.
Currently only linux-64/python3.6 conda build is available:
conda install -c fastai -c fastai/label/test fastai torchvision=0.2.1=pyhe7f20fa_0
We had to build a special version of torchvision which depends on pytorch-nightly.
For other setups/platforms use pip install at the moment (see above).
conda install pytorch-nightly -c pytorch
conda install torchvision -c pytorch
git clone https://github.com/fastai/fastai_pytorch
cd fastai_pytorch
pip install -e .
tools/run-after-git-clone
Please refer to CONTRIBUTING.md and the developers guide for more details.