Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Dense prompt following on DPG-Bench v1.0 (test)

93.08Entity Score

Nucleus-Image

73.47678.565583.65588.7445Apr 14, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
93.0885.192.293.5693.6288.79
92.6594.3191.3692.7888.2488.27
91.97-90.294.85-87.2
2026.04
91.5691.3292.0294.3192.7388.32
2026.04
91.0187.988.8380.788.6884.08
90.2276.4489.4893.7491.8385.89
2026.04
9074.3588.9690.8788.3383.84
2026.04
89.6190.9788.3990.5889.8383.5
88.9488.8989.8492.6390.9685.15
88.986.989.489.3289.4884.19
88.6582.8286.4480.5381.8274.63
88.6387.5888.1788.9888.382.63
2026.04
87.3882.3387.755.4686.4179.68
2026.04
86.6885.2186.8490.2283.1580.6
2026.04
82.8986.8988.9486.5987.6880.54
82.5983.0681.284.0883.575.47
2026.04
82.4383.2780.9186.7680.4174.65
2026.04
80.5984.5988.0174.3686.4178.87
2026.04
79.3274.9778.682.5776.9671.11
2026.04
74.2374.6375.3973.4967.8163.18