Multiple Choice Question Answering on CommonsenseQA (CQA)
85.09AccuracyQwen3 30B
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Qwen3 30Bconfidence_type=normalized, evaluation_protocol=output probabilities2026.02 | 85.09 | 84.7 | 12.4 | |
| GLM4 9Bconfidence_type=normalized, evaluation_protocol=output probabilities2026.02 | 81.82 | 83.3 | 2.5 | |
| Gemma3 12Bconfidence_type=normalized, evaluation_protocol=output probabilities2026.02 | 78.71 | 82 | 20 | |
| Qwen3 4Bconfidence_type=normalized, evaluation_protocol=output probabilities2026.02 | 78.21 | 80.3 | 19.6 | |
| Gemma3 4Bconfidence_type=normalized, evaluation_protocol=output probabilities2026.02 | 69.7 | 75.6 | 29.5 |