Mathematical Reasoning on GSM8K MINCE Calibration n=400
74.5AccuracyPhi-4-mini-Inst
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Phi-4-mini-InstPrecision=BF162026.06 | 74.5 | 1.16 | |
| Llama-3.1-8B-InstPrecision=BF162026.06 | 71 | 2.54 | |
| Qwen3-8BPrecision=BF162026.06 | 55.25 | 0.85 | |
| Qwen3.5-35B-A3BPrecision=BF162026.06 | 54 | 0.44 | |
| Qwen2.5-3B-InstPrecision=BF162026.06 | 44 | 1.56 | |
| Mistral-7B-InstPrecision=BF162026.06 | 40.75 | 2.62 | |
| Llama-3.1-8B (base)Precision=BF162026.06 | 25 | 0.63 |