List your company

Deep Research Evaluation on Deep Research Bench (step 1100)

53.81ReadabilityTournament-GRPO
32.261237.855643.4549.0444May 26, 2026

Evaluation results

MethodLinks
Tournament-GRPOGroup=G = 4, Kwin = 2, Kfinal = 2, M=82026.0553.8155.2255.1955.455.09
Tournament-GRPOGroup=G = 2, Kwin = 1, Kfinal = 1, M=42026.0553.6954.9255.6355.1655.04
Tournament-GRPOGroup=G = 4, Kwin = 2, Kfinal = 2, M=22026.0553.0353.5453.8353.853.64
Tournament-GRPOGroup=G = 2, Kwin = 1, Kfinal = 1, M=82026.0552.9353.953.9654.0453.84
Tournament-GRPOGroup=G = 4, Kwin = 2, Kfinal = 2, M=12026.0552.7953.2652.9553.3853.14
Tournament-GRPOGroup=G = 2, Kwin = 1, Kfinal = 1, M=22026.0552.3853.0453.1752.9152.96
PairwiseGroup=Baseline2026.0551.1551.3450.3850.0550.57
GDPOGroup=Baseline2026.0551.1151.0150.5250.5250.71
Tournament-GRPOGroup=G = 2, Kwin = 1, Kfinal = 1, M=12026.0550.0650.7650.1649.9450.2
Tournament-GRPOGroup=G = 4, Kwin = 2, Kfinal = 2, M=42026.0549.6550.9250.5350.0150.32
Tournament-GRPOGroup=G = 2, Kwin = 1, Kfinal = 1, M=32026.0548.6949.4347.9446.2447.81
Tournament-GRPOGroup=G = 4, Kwin = 2, Kfinal = 2, M=32026.0545.684646.2346.0146.04
GRPO explicitGroup=Baseline2026.0541.2641.5641.3340.7741.2
Dr. GRPO implicitGroup=Baseline2026.0540.8838.9236.4235.4737.38
Dr. GRPO explicitGroup=Baseline2026.0540.4938.9836.436.0737.51
GRPO implicitGroup=Baseline2026.0533.0932.9732.1431.1732.14