yangbobo2021/tiny-arith-gym
Take an 800K-param GPT from 2.2% to 95.5% on arithmetic through 20 controlled experiments — a hands-on tour of LLM training methods, including the ones that failed.
Python
2
MIT License
Social mentions
Recent discussions about this repository across the web
Repository activities
repository's daily and monthly activities across stars, forks, merged PRs, issues, and closed issues