llm.c
llm.c copied to clipboard
added reading checkpoint files
This refers to reading checkpoints in HF format (issue].
In the spirit of readability, I have tried to keep the code as close as possible to train_gpt2.py.
I have also included tests, which can be extended to models different from GPT-2 family.
Note: I am not a software engineer and this happens to be my first ever pull request to a public repo, so expect errors and omissions :)