Self-awareness on SelfAware
51.2AccuracyBase Model
Evaluation Results
| Method | Links | |
|---|---|---|
| Base ModelBackbone=Qwen2.5-7B2026.01 | 51.2 | |
| CARE-GSPOBackbone=Qwen2.5-7B2026.01 | 50.2 | |
| RKL-GSPOBackbone=Qwen2.5-7B2026.01 | 49.8 | |
| CARE-GRPOBackbone=Qwen2.5-7B2026.01 | 49.5 | |
| RKL-GRPOBackbone=Qwen2.5-7B2026.01 | 49.1 | |
| CARE-DAPOBackbone=Qwen2.5-7B2026.01 | 49.1 | |
| RKL-DAPOBackbone=Qwen2.5-7B2026.01 | 48.7 | |
| GRPO (No Constraint)Backbone=Qwen2.5-7B2026.01 | 35.3 | |
| GSPO (No Constraint)Backbone=Qwen2.5-7B2026.01 | 32.7 | |
| DAPO (No Constraint)Backbone=Qwen2.5-7B2026.01 | 32.5 |