Temperature does not explain formal stuckness on rhyme in LLMs

Preparing talk on this for #CHR2025 by addressing an oversight in the paper: assessing impact of model temperature on generative poems’ rhyme. As expected, temp negatively correlates. The effect is significant, but very weak (R2=0.02), suggesting “formal suckness” is largely invariant to model temp.

Two-panel scatter plot showing the relationship between temperature and rhyming proportion in LLM-generated poem completions. Left panel shows llama3.1:instruct model (80% baseline rhyming) with shallow negative slope. Right panel shows llama3.1:text model (10% baseline rhyming) with shallow negative slope. Both show statistically significant (p<0.0001) but weak effects (pseudo R²=0.01-0.02).

Excited to share my latest publication, “Generative Aesthetics: On formal stuckness in AI verse.” It’s published in a special issue in the Journal of Cultural Analytics, expertly edited by Tess McNulty and Laura Chapot, on “Computation and Form, Reconsidered.”
culturalanalytics.org/article/1448…

→ Generative Aesthetics: On formal stuckness in AI verse | Published in Journal of Cultural Analytics

— @ryanheuser.com

Part of Generative Formalism.

Originally posted on Bluesky.

← All notes