1 paper
Jungsoo Park, Hyungjoo Chae, Ethan Mendes +4
Large language models can predict real-valued quantities from heterogeneous inputs such as text, code, and molecular strings, but most training objectives score each decoded floati…