This is a small language model (~60M params) trained from scratch using NanoGPT-style architecture on:
This model is intended for research and learning, not production.
-