set_up_environment.md

Setting Up Your Environment

This project uses a conda environment to manage dependencies. If you need conda, you can get it from Miniconda or any of the other Anaconda bundles. Once conda is set up, enter the requirements directory and run

$ conda create ladi-v2-env -y --file conda_requirements -c pytorch -c nvidia
$ conda activate ladi-v2-env
$ pip install -r pip_requirements

This will create an environment called ladi-v2-env and can be activated with conda activate ladi-v2-env

Exact environment

We have also provided an environment.yml file to document the exact configuration used to generate our results. This may be useful for debugging package versions. The environment was created on an Intel-based system with Nvidia graphics running Ubuntu 22.04 (Jammy), and may not reproduce if your system configuration differs. To try and install the exact reproduction environment, you can run.

conda env create -f environment.yml

The environment's name, defined in the yaml file, is ladi-v2-env and can be activated with conda activate ladi-v2-env.