List your company

Science Question Answering on GPQA Diamond (Score)

70ScoreBaseline
66.56867.45968.3569.241Jul 5, 2026

Evaluation results

MethodLinks
BaselinePrecision=BF162026.0770
AFM-k5984d3s (Ours)Optimization=QAD + Speculative Decoding + SWA2026.0766.7