Installing coreai-torch

Set up a Python environment and install coreai-torch.

Overview

This article covers two installation paths for coreai-torch: uv and conda. Use uv when possible — it creates an isolated virtual environment, pins the correct Python version, and resolves all dependencies in one command.

Prerequisites

• Python 3.11 or later

• PyTorch 2.8.0 or later

Install with uv

uv is the recommended package manager. For a development environment, run uv sync from the repository root:

uv sync

This installs standard dependencies in editable mode. To also include test dependencies, run:

uv sync --extra test

Install with conda

If you prefer conda, create and activate a new environment first:

conda create -n coreai-torch python=3.11 -y
conda activate coreai-torch

Then install from source — clone the repository, navigate to its root, and run:

pip install -e .

Verify installation

Run the following to confirm coreai-torch is installed correctly — a version string confirms success:

import coreai_torch
print(coreai_torch.__version__)

Next steps

Head to the Quickstart tutorial to convert your first PyTorch model and explore more.

Notices

PyTorch is a trademark of Meta Platforms, Inc. conda is a product of Anaconda, Inc.