Correct Choices, Failed Sequences: Choice2Seq Benchmarks Correctness Retention Across Output Interfaces
A model may select the correct answer yet fail to construct it through a sequence of local decisions. We introduce Choice2Seq, a diagnostic benchmark that measures correctness retention across complete-answer selection and sequential construction in independent runs on the same problems and candidate sets. Sequence Com...
Tian-You Lai, Wen-Tao Yue, Zhi-Ying Deng et al.
· 0 citations