General Reasoning on LB-R
97AccuracyCOSE
Evaluation results
| Method | Links | |
|---|---|---|
| COSEModel=Qwen3-0.6B2026.05 | 97 | |
| COSEModel=Qwen3-4B2026.05 | 94.41 | |
| COSEModel=Llama-3.2-3B-Instruct2026.05 | 93.51 | |
| MAEModel=Llama-3.2-3B-Instruct2026.05 | 93 | |
| MAEModel=Qwen3-4B2026.05 | 79.4 | |
| AZRModel=Qwen3-4B2026.05 | 78.8 | |
| R-ZeroModel=Qwen3-4B2026.05 | 78.1 | |
| BaseModel=Qwen3-4B2026.05 | 75 | |
| R-ZeroModel=Llama-3.2-3B-Instruct2026.05 | 36.56 | |
| AZRModel=Llama-3.2-3B-Instruct2026.05 | 34.12 | |
| BaseModel=Llama-3.2-3B-Instruct2026.05 | 32 | |
| MAEModel=Qwen3-0.6B2026.05 | 31.4 | |
| R-ZeroModel=Qwen3-0.6B2026.05 | 30.2 | |
| COSEModel=Qwen2.5-3B-Instruct2026.05 | 28.8 | |
| AZRModel=Qwen3-0.6B2026.05 | 28.22 | |
| BaseModel=Qwen3-0.6B2026.05 | 26 | |
| MAEModel=Qwen2.5-3B-Instruct2026.05 | 23.48 | |
| R-ZeroModel=Qwen2.5-3B-Instruct2026.05 | 22.9 | |
| AZRModel=Qwen2.5-3B-Instruct2026.05 | 22.27 | |
| BaseModel=Qwen2.5-3B-Instruct2026.05 | 20.8 |