Mr. Chatterbox is a Victorian-era ethically trained model
simonwillison.net
March 31, 2026
4 min read
🔥🔥🔥🔥🔥
48/100
Summary
Mr. Chatterbox is a language model trained on over 28,000 Victorian-era British texts published between 1837 and 1899. The model can be run locally on personal computers and is based on a dataset provided by the British Library.
Key Takeaways
- Mr. Chatterbox is a language model trained on over 28,000 Victorian-era British texts published between 1837 and 1899, using a dataset from the British Library.
- The model has approximately 340 million parameters and is trained entirely on historical data, with no inputs from after 1899.
- Mr. Chatterbox is relatively small at 2.05GB and is described as providing responses that resemble a Markov chain rather than a sophisticated conversational partner.
- The training corpus consists of about 2.93 billion tokens, which is significantly less than the amount suggested for more effective language models.
Community Sentiment
Positives
- The model's ethical training approach ensures compliance with copyright laws, which is crucial for responsible AI development and deployment.
- Using this model could help demystify LLMs for the general public, providing a clearer understanding of how text prediction works.
Concerns
- The model's 340 million parameters seem insufficient for generating coherent Victorian speech, raising concerns about its effectiveness in capturing the desired style.
- Limiting training data to works published before 1899 may overlook significant copyright issues, potentially impacting the model's applicability and reliability.
