Efficient semantic uncertainty quantification in language models via diversity-steered sampling

Open in new window