List your company

Mathematical Reasoning on MATHINSTRUCT Evaluation Suite (In-domain and Out-of-domain)

54.89GSM8K AccuracyUniPROT
20.372429.333738.29547.2563Apr 13, 2026

Evaluation results

MethodLinks
UniPROTBackbone=ZEPHYR-3B, Selection Configuration=batch-wise, Batch Size=128, Total Budget=322026.0454.8919.640.438.554.115.522.9531.3434.92
UniPROTBackbone=ZEPHYR-3B, Selection Configuration=source-wise, Batch Size=128, Total Budget=322026.0454.420.441.338.554.316.524.731.3434.92
GREATSBackbone=ZEPHYR-3B, Selection Configuration=source-wise, Batch Size=128, Total Budget=322026.0452.819.0140.836.3254.517.121.9830.6433.48
GradNormBackbone=ZEPHYR-3B, Selection Configuration=source-wise, Batch Size=128, Total Budget=322026.0452.316.54035.753.515.522.530.4533.07
FTBackbone=ZEPHYR-3B, Batch Size=128, Total Budget=322026.0452.0816.1440.136.1154.715.622.230.8333.47
SBERTBackbone=ZEPHYR-3B, Selection Configuration=source-wise, Batch Size=128, Total Budget=322026.0451.314.641.628.4456.316.221.9826.0627.25
COLMBackbone=ZEPHYR-3B, Selection Configuration=source-wise, Batch Size=128, Total Budget=322026.0450.621.4240.1536.5655.816.722.1730.8633.71
GREATSBackbone=ZEPHYR-3B, Selection Configuration=batch-wise, Batch Size=128, Total Budget=322026.0450.615.938.836.3254.613.921.7830.6433.48
COLMBackbone=ZEPHYR-3B, Selection Configuration=batch-wise, Batch Size=128, Total Budget=322026.0449.8817.439.936.5654.115.421.0130.8633.71
GradNormBackbone=ZEPHYR-3B, Selection Configuration=batch-wise, Batch Size=128, Total Budget=322026.0449.815.739.935.753.115.422.730.4533.07
SBERTBackbone=ZEPHYR-3B, Selection Configuration=batch-wise, Batch Size=128, Total Budget=322026.0421.714.0127.4428.4434.513.613.826.0627.25