List your company

Multi-turn Dialogue on Honor-Dialogue

88.13Life Services Domain PerformanceDVPO
27.70643.39359.0874.767Dec 3, 2025

Evaluation results

MethodLinks
DVPOtraining steps=5002025.1288.1387.7387.6787.6782.7386.79
Reinforce++training steps=5002025.1287.8781.0777.6787.1385.5383.85
Baselinetraining steps=5002025.1286.7384.590.2387.1382.786.26
PPOtraining steps=5002025.1284.286.0790.878183.8785.2
Dr.GRPOtraining steps=5002025.1251.6760.361.3752.1756.4756.39
GRPOtraining steps=5002025.1230.0328.5728.3328.927.928.75