Skip to main content

Fast and customizable framework for automatic ML model creation (AutoML)

Project description

Developing LightAutoML on GPU

To develop LightAutoML on GPUs using RAPIDS some prerequisites need to be met:

  1. NVIDIA GPU: Pascal or higher
  2. CUDA 11.0 (drivers v460.32+) or higher need to be installed
  3. Python version 3.8 or higher
  4. OS: Ubuntu 16.04/18.04/20.04 or CentOS 7/8 with gcc/++ 9.0+

Installation

Anaconda or Miniconda is necessary to install RAPIDS and work with environments.

  1. Once you install Anaconda/Miniconda, you need to set up your own environment. For example:
conda create -n lama_venv python=3.8
conda activate lama_venv
  1. To install RAPIDS for Python 3.8 and CUDA 11.0 use the following command:
conda install -c rapidsai -c nvidia -c conda-forge rapids=22.10 cudatoolkit=11.0
pip install dask-ml
  1. To clone the project on your own local machine:
git clone https://github.com/ekonyagin/LightAutoML-1.git
cd LightAutoML-1
  1. Install LightAutoML in develop mode and other necessary libraries:
pip install .
pip install catboost
pip install py-boost

After you change the library code, you need to re-install the library: go to LightAutoML directory and call pip install ./ -U

Please note, if you use NVIDIA GPU Ampere architecture (i.e. Tesla A100 or RTX3000 series), you may need to uninstall pytorch and install it manually due to compatibility issues. To do so, run following commands:

pip uninstall torch torchvision
pip install torch==1.7.1+cu110 torchvision==0.8.2+cu110 -f https://download.pytorch.org/whl/torch_stable.html

Once the RAPIDS is installed, the environment is fully ready. You can activate it using the source command and test and implement your own code.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

lightautoml-gpu-0.2.15.tar.gz (349.3 kB view details)

Uploaded Source

Built Distribution

lightautoml_gpu-0.2.15-py3-none-any.whl (500.3 kB view details)

Uploaded Python 3

File details

Details for the file lightautoml-gpu-0.2.15.tar.gz.

File metadata

  • Download URL: lightautoml-gpu-0.2.15.tar.gz
  • Upload date:
  • Size: 349.3 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: poetry/1.0.9 CPython/3.6.9 Linux/4.15.0-197-generic

File hashes

Hashes for lightautoml-gpu-0.2.15.tar.gz
Algorithm Hash digest
SHA256 f3f71b51f1d68262965155f04a74e31b09c74ae34e9bdb9a2d24c77ac01abb3e
MD5 8474a9fefcd67c25eca9c72c31562f9b
BLAKE2b-256 ca851b8341b92fa6accdcbabf90d5a9cc40093aaf1dfa4f8c560f65e795fa843

See more details on using hashes here.

File details

Details for the file lightautoml_gpu-0.2.15-py3-none-any.whl.

File metadata

  • Download URL: lightautoml_gpu-0.2.15-py3-none-any.whl
  • Upload date:
  • Size: 500.3 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: poetry/1.0.9 CPython/3.6.9 Linux/4.15.0-197-generic

File hashes

Hashes for lightautoml_gpu-0.2.15-py3-none-any.whl
Algorithm Hash digest
SHA256 00f7a6c06b9ed713e4210099627ed5fa0ade8043bc030655d92287d70ce9b9a3
MD5 3c4d66225fe740f3bf00d31d686e9417
BLAKE2b-256 9cf663b7030953f7dc58de9d63af8adfe11a3a7ffee76a5775ceb51672ec1d04

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page