An Artist Trains an LLM on Her Own Writing

Merzmensch · x · 2026-07-11

The author explores moving beyond using AI merely as a tool to training an LLM herself. The model was trained on 20 years of her personal work, encompassing poetry, essays, plays, criticism, and books.

She describes the dataset processing as highly revealing, even experimenting with training a version exclusively on her MERZAZINE texts. The goal is to empower artists to build custom creative tools on their own terms, free from reliance on proprietary systems.

Related event: Artist Trains Custom LLM on 20 Years of Personal Writing(3 posts)→

Original post →

More from Apps

Apps channel →