General Reasoning on TriviaQA
73.33AccuracyCOSE
Evaluation Results
| Method | Links | |
|---|---|---|
| COSEModel=Qwen3-4B2026.05 | 73.33 | |
| COSEModel=Llama-3.2-3B-Instruct2026.05 | 71.86 | |
| COSEModel=Qwen3-0.6B2026.05 | 71.55 | |
| MAEModel=Qwen3-4B2026.05 | 68.9 | |
| AZRModel=Qwen3-4B2026.05 | 68.4 | |
| R-ZeroModel=Qwen3-4B2026.05 | 68.2 | |
| BaseModel=Qwen3-4B2026.05 | 66 | |
| MAEModel=Qwen2.5-3B-Instruct2026.05 | 60.13 | |
| R-ZeroModel=Llama-3.2-3B-Instruct2026.05 | 59.53 | |
| MAEModel=Llama-3.2-3B-Instruct2026.05 | 59 | |
| AZRModel=Llama-3.2-3B-Instruct2026.05 | 58.23 | |
| R-ZeroModel=Qwen2.5-3B-Instruct2026.05 | 58.2 | |
| BaseModel=Llama-3.2-3B-Instruct2026.05 | 58 | |
| AZRModel=Qwen2.5-3B-Instruct2026.05 | 56.67 | |
| COSEModel=Qwen2.5-3B-Instruct2026.05 | 53 | |
| BaseModel=Qwen2.5-3B-Instruct2026.05 | 51.6 | |
| MAEModel=Qwen3-0.6B2026.05 | 25.8 | |
| R-ZeroModel=Qwen3-0.6B2026.05 | 24.6 | |
| AZRModel=Qwen3-0.6B2026.05 | 21.75 | |
| BaseModel=Qwen3-0.6B2026.05 | 18 |