Now that you have some code in the repo it is a good idea to create Python virtual environment files that make the necessary dependencies for those environments reproducible. So this would be something like a requirements.txt file that you can use with pip or an environment.yml file that you can use with Conda (If you're using Conda ideally you'd have both so that if someone wants to get your environment without using Conda they can).
Now that you have some code in the repo it is a good idea to create Python virtual environment files that make the necessary dependencies for those environments reproducible. So this would be something like a
requirements.txtfile that you can use withpipor anenvironment.ymlfile that you can use with Conda (If you're using Conda ideally you'd have both so that if someone wants to get your environment without using Conda they can).So from
https://github.com/mihirkatare/DeepMEM/blob/9b33d4e179d062d37edce1007272e5ff413999bf/debug/debugdata.py#L1-L4
we can see that this (at minimum) would require something like a
requirements.txtfile containingas
pandaswill pickup the required version ofnumpyandtorchvisionwill gettorch