Factuality on SelfAware
0.372ScoreBase Model
Evaluation Results
| Method | Links | |
|---|---|---|
| Base ModelBackbone=Qwen2.5-3B2026.01 | 0.372 | |
| CARE-GRPOBackbone=Qwen2.5-3B, Algorithm=GRPO, Constraint=CARE2026.01 | 0.355 | |
| RKL-GRPOBackbone=Qwen2.5-3B, Algorithm=GRPO, Constraint=RKL2026.01 | 0.351 | |
| RKL-DAPOBackbone=Qwen2.5-3B, Algorithm=DAPO, Constraint=RKL2026.01 | 0.346 | |
| RKL-GSPOBackbone=Qwen2.5-3B, Algorithm=GSPO, Constraint=RKL2026.01 | 0.341 | |
| CARE-DAPOBackbone=Qwen2.5-3B, Algorithm=DAPO, Constraint=CARE2026.01 | 0.334 | |
| CARE-GSPOBackbone=Qwen2.5-3B, Algorithm=GSPO, Constraint=CARE2026.01 | 0.332 | |
| GRPO (No Constraint)Backbone=Qwen2.5-3B, Algorithm=GRPO, Constraint=None2026.01 | 0.249 | |
| GSPO (No Constraint)Backbone=Qwen2.5-3B, Algorithm=GSPO, Constraint=None2026.01 | 0.243 | |
| DAPO (No Constraint)Backbone=Qwen2.5-3B, Algorithm=DAPO, Constraint=None2026.01 | 0.232 |