Mr. Chatterbox: Proof of Concept for Training LLMs Without Web Scraping

Mr. Chatterbox demonstrates that viable language models can be trained using only public domain texts, offering a copyright-free alternative to internet scraping. While the resulting model is notably weak, the proof-of-concept suggests a path forward for ethical LLM development without legal concerns. The tiny 340M-parameter model also proves that useful AI can run locally on consumer hardware.