works on
evaluation metrics 1generalization across domains 1ideological bias 1language model finetuning 1model alignment 1

From the 1 of 3 linked papers with an AI index.

collaborators
Showing cs.LGShow all

1 paper · 1 filter

cs.LG2026

Innocuous-Seeming Data, Latent Ideology: Ideological Generalisation in Finetuned LLMs

Robert Graham, Edward Stevinson, Yariv Barsheshat

The paper shows that fine‑tuning large language models on small, factually defensible datasets can cause broad ideological shifts across unrelated topics, and introduces metrics to…