Direct Preference Optimization on UltraFeedback
69.92AccuracyFedPDPO
Evaluation Results
| Method | Links | |
|---|---|---|
| FedPDPODomain Setting (Cross-Domain)=Cross-Domain, Optimization Objective (PPO/DPO)=DPO2026.03 | 69.92 | |
| FedPerDomain Setting (Cross-Domain)=Cross-Domain, Optimization Objective (PPO/DPO)=DPO2026.03 | 68.02 | |
| Per-FedAvgDomain Setting (Cross-Domain)=Cross-Domain, Optimization Objective (PPO/DPO)=DPO2026.03 | 67.48 | |
| FedRepDomain Setting (Cross-Domain)=Cross-Domain, Optimization Objective (PPO/DPO)=DPO2026.03 | 67.25 | |
| FedAvgDomain Setting (Cross-Domain)=Cross-Domain, Optimization Objective (PPO/DPO)=DPO2026.03 | 66.45 | |
| FedRepDomain Setting (Cross-Domain)=Cross-Domain, Optimization Objective (PPO/DPO)=PPO2026.03 | 65.54 | |
| FedAMPDomain Setting (Cross-Domain)=Cross-Domain, Optimization Objective (PPO/DPO)=DPO2026.03 | 65.47 | |
| FedPerDomain Setting (Cross-Domain)=Cross-Domain, Optimization Objective (PPO/DPO)=PPO2026.03 | 65.45 | |
| Per-FedAvgDomain Setting (Cross-Domain)=Cross-Domain, Optimization Objective (PPO/DPO)=PPO2026.03 | 64.67 | |
| FedAMPDomain Setting (Cross-Domain)=Cross-Domain, Optimization Objective (PPO/DPO)=PPO2026.03 | 63.98 | |
| FedAvgDomain Setting (Cross-Domain)=Cross-Domain, Optimization Objective (PPO/DPO)=PPO2026.03 | 63.89 |