ResearchBenchmarksResponse Generation on DecTest resp_gen no_hds (1000 samples)Follow0.924Spearman ρa_n-0.88248-0.413490.05550.52449Jun 1, 2026Evaluation ResultsMethodMethodLinksSpearman ρa_nmodel=Qwen2.5-3B, perm...model=Qwen2.5-3B, permutations=252026.060.924cos-sim2026.060.895distinct-n2026.060.894BERTScore2026.060.874SentBERT2026.060.801C x a_nmodel=Qwen2.5-3B, perm...model=Qwen2.5-3B, permutations=252026.060.771Cmodel=Qwen2.5-3B, perm...model=Qwen2.5-3B, permutations=252026.06-0.813