nanoGPT icon indicating copy to clipboard operation
nanoGPT copied to clipboard

Requirements & encoding

Open nil-andreu opened this issue 2 years ago • 3 comments

This is a small contribution on some things that I found interesting to take into consideration:

  • requirements.txt: provide an easy way to create the environment.
  • .gitignore: ignore the virtual environment as well as the data files
  • encoding: specify the encoding with the I/O operations to make sure that the data is consistent across the Operating Systems

For me is totally fine if you want to review/discuss/reject them. Let me know if it is needed anything else.

nil-andreu avatar Jan 03 '23 20:01 nil-andreu

Hi ty

  • I like the encodings change
  • I don't like boilerplate gitignores, I like when they're specific to repo
  • pip installing pytorch would not grab the nightly I believe?

karpathy avatar Jan 03 '23 21:01 karpathy

Hi ty Hello!

  • I like the encodings change Thankss! Fine
  • I don't like boilerplate gitignores, I like when they're specific to repo Okay, could remove the things that are general and make it only specific to a repository.
  • pip installing pytorch would not grab the nightly I believe? Not, but I saw the issue that this package was having with the 32 precision. So I could add the nightly as well in the requirements.

I would like to be contributor if you need any help on more development!

nil-andreu avatar Jan 04 '23 06:01 nil-andreu

  • pip installing pytorch would not grab the nightly I believe?

You can point it at the nightly index URLs, but to do so via pip requires choosing a CUDA version to target as they have different URLs. This is a better dev experience for those who have that version, but worse for everyone else who doesn't.

bradezard131 avatar Jan 13 '23 00:01 bradezard131

I'd like to have better requirements flow for the repo but don't love this specific approach. Will think on. Ty.

karpathy avatar Jan 19 '23 23:01 karpathy