Skip to content

Latest commit

 

History

History
24 lines (19 loc) · 1.52 KB

INSTALL.md

File metadata and controls

24 lines (19 loc) · 1.52 KB

To use the Pytorch Experiment Template, one needs to set up a conda environment with the right dependencies. This can be achieved as follows:

  1. Point your web-browser to miniforge

  2. Find and download a suitable mini forge installer for your OS. The next few steps assume Linux and MacOS

  3. Go to the directory in which you have downloaded the minforge installer and run:bash <mini-forge installer>

  4. Follow the setup to install the environment.

  5. Once done, activate conda using conda activate, note that you might need to resource your .bashrc using source ~/.bashrc to get access to conda if you are using the same terminal as the one used to install conda.

  6. Create a new conda environment using conda create -n pytorch-experiments python=3.7

  7. Once done, run conda install git tqdm rich matplotlib regex ftfy requests seaborn GPUtil scipy; Follow through the process to finish installing these packages.

8.Then go to the Pytorch installation page and run the appropriate commands to install pytorch on your machine. The command is not given here as it depends on your OS, CPU/GPU type and CUDA version (if any). This one will take a while.

Note: PyTorch >= 1.8.0 is required for proper usage of this codebase

  1. You should now be able to run the default resnet18, cifar10 experiment using python train.py.

If you have any problems or further questions please post on the slack channel "#pytorch-debug"