nanochat-d32 / README.md
karpathy's picture
Update README.md
016dba0 verified
|
raw
history blame
526 Bytes
metadata
license: mit

The nanochat-d32 model described in detail here.

I'm sorry this is a janky upload but you have to place these files correctly on your end:

  • the token_bytes.pt, tokenizer.pkl have to go into ~/.cache/nanochat/tokenizer directory
  • the meta_000650.json, model_000650.pt have to go into ~/.cache/nanochat/chatsft_checkpoints/d32/

I'll figure out how to make this less janky in the future, and to make nanochat play nicer with huggingface infra.