List your company

Question Answering on Natural Questions (NQ) (Exact Match Metrics)

58SubEMDSPO
35.43241.29147.1553.009Oct 10, 2025

Evaluation results

MethodLinks
DSPOModel Scale=7B, Dynamic Filter=true, Sequence-level Optimization=true2025.1058—
Search-R1 (GRPO, 14B)Model Scale=14B, RL Algorithm=GRPO2025.1053.548.2
DSPO w/o seq-lvl opt.Model Scale=7B, Dynamic Filter=true, Sequence-level Optimization=false2025.1047—
Search-R1 (GRPO, 7B)Model Scale=7B, RL Algorithm=GRPO2025.1042.342.9
DSPO w/o dyn. filterModel Scale=7B, Dynamic Filter=false, Sequence-level Optimization=true2025.1036.3—
Search-R1 (PPO, 14B)Model Scale=14B, RL Algorithm=PPO2025.10—42.4
Search-R1 (PPO, 7B)Model Scale=7B, RL Algorithm=PPO2025.10—39.3