Tag
This paper investigates how domain adaptation reshapes explanatory behavior in language models by training on a pre-Copernican corpus, finding that fine-tuning shifts explanatory framing more than cosmological stance.