Preprint
Aug 2026
Predicting LLM Performance from Prompt Linguistic Features: An Empirical Study in Requirements Engineering
Whether measurable linguistic properties of prompts can predict LLM performance before inference is investigated, enabling low-cost prompt selection and refinement, validated on binary requirements classification targeting F1, F2, precision, and recall.
Quim Motger, Alessio Miaschi, Xavier Franch et al.
· 1 citation