@RyanMorey
@emollick Found a bit more detail from the creator here! Trained with @karpathy ’s Nanochat https://t.co/TCLYtWCPIO “Using nanochat, I built a small LLM experiment called Mr. Chatterbox, a chatbot trained entirely on books published during the Victorian era (1837–1899). It was trained on a subset of the BL Books dataset, then fine-tuned on a mix of corpus and synthetic data. I used nanochat for the initial training and supervised fine-tuning rounds. SFT consisted of two rounds: one round of two epochs on a large dataset (over 40,000 pairs) of corpus material and synthetic data, and a smaller round that focused on specific cases like handling modern greetings, goodbyes, attempted prompt injections, etc.”