Mathematical Reasoning on MathQA (Accuracy)
43.45AccuracyFP16
Evaluation results
| Method | Links | |
|---|---|---|
| FP16BIT=16, Evaluation Protocol=Few-shot2026.07 | 43.45 | |
| GSRQBIT=2, Evaluation Protocol=Few-shot2026.07 | 41.07 | |
| VQLLMBIT=2, Evaluation Protocol=Few-shot2026.07 | 39.16 | |
| GSRQBIT=1, Evaluation Protocol=Few-shot2026.07 | 34.84 | |
| GSRQBIT=0.75, Evaluation Protocol=Few-shot2026.07 | 31.76 | |
| VQLLMBIT=1, Evaluation Protocol=Few-shot2026.07 | 25.23 |