yuchenj's picture
Update README.md
0fc8efb verified
metadata
library_name: transformers
datasets:
  - HuggingFaceFW/fineweb-edu
license: mit

This is a GPT-2 (350M) model trained in llm.c for 100B tokens with WSD (Warmup-Stable-Decay) learning rate schedule on FineWeb-EDU.

A lot more detailed info and observations are here: https://x.com/Yuchenj_UW/status/1816508452518482319