High-level abstractions for PyTorch.
BxTorch is a high-level library for working with PyTorch. It is designed to make PyTorch much simpler in the most common cases. Yet, it is engineered to be highly extensible in order to preserve PyTorch's flexibility --- while relieving you from writing boilerplate code.
BxTorch is available on PyPi, so simply run the following command:
pip install bxtorch
The package will install the dependencies specified here. If you plan to use plotting features of BxTorch, make sure to also install the following packages:
Generally, BxTorch provides an object-oriented approach to abstracting PyTorch's API. The core design objective is to provide an API both as simple and as extensible as possible --- usually at the expense of some milliseconds of execution time. Be aware that the goal of this library is not to maximize performance in cases where it is not needed.
This does not mean that BxTorch does not care about performance: in fact, the library has built-in support for multi-GPU training, both within a single process and split over multiple processes.
It must be emphasized that BxTorch is not meant to be a wrapper for PyTorch as Keras is for TensorFlow, for example. It only provides extensions for PyTorch.
BxTorch is licensed under the MIT License.
The logo is modified from thenounproject.com, "Torch by iconsmind.com from the Noun Project".
Release history Release notifications
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
|Filename, size||File type||Python version||Upload date||Hashes|
|Filename, size BxTorch-0.2.40-py3-none-any.whl (59.1 kB)||File type Wheel||Python version py3||Upload date||Hashes View hashes|
|Filename, size BxTorch-0.2.40.tar.gz (38.7 kB)||File type Source||Python version None||Upload date||Hashes View hashes|