Installation

We recommend installation of PGBM in a (virtual) Python environment via PyPi.

Don’t know how to setup a Python environment and/or access a terminal? We recommend you start by installing Anaconda. This allows you to create Python virtual environments and access them via a terminal.

Installing via PyPi

First, make sure you have separately installed the various dependencies (see below). Then, execute the following commands in a terminal after you have activated your Python (virtual) environment.

pip install pgbm

Installing from source

Clone the repository and run setup.py install in the newly created directory pgbm, i.e. run the following code from a terminal within a Python (virtual) environment of your choice:

git clone https://github.com/elephaint/pgbm.git
cd pgbm
python setup.py install

When installing from source you need to install the dependencies separately.

Dependencies

We offer PGBM using two backends, PyTorch (import pgbm.torch) and scikit-learn (import pgbm.sklearn). Only the PyTorch-backend requires you to install dependencies separately.

Torch backend

torch>=1.8.0, with CUDA Toolkit >= 10.2 for GPU acceleration. Verify that PyTorch can find a cuda device on your machine by checking whether torch.cuda.is_available() returns True after installing PyTorch.
GPU training: the CUDA device should have CUDA compute ability 2.x or higher.
Windows users: you may need to install Build Tools for Visual Studio and make sure you add compiler cl to your PATH environment variable (see here). Verify that Windows can find cl by executing where cl in a Windows command line terminal.

Verification

Both backends use JIT-compilation so you incur additional compilation time the first time you use PGBM. The compiled code will be subsequently cached for all versions.

To verify, download & run an example from the examples folder to verify the installation is correct:

Run this example to verify ability to train & predict on CPU with Torch backend.
Run this example to verify ability to train & predict on GPU with Torch backend.
Run this example to verify ability to train & predict on CPU with Scikit-learn backend.
Run this example to verify ability to perform distributed CPU, GPU, multi-CPU and/or multi-GPU training.