List your company

Mathematical Reasoning on GSM8K (Performance vs Context Size)

74.3Acc (128)EvoToken (10k FT)
48.0454.857561.67568.4925Jan 12, 2026

Evaluation results

MethodLinks
EvoToken (10k FT)NFE / Gen Len=12026.0174.383.4784.4680.74
EvoToken (10k FT)NFE / Gen Len=1/22026.0173.5482.0381.879.12
FT-Base (10k FT)NFE / Gen Len=12026.0171.0482.1182.5678.57
BaselineNFE / Gen Len=12026.0170.279.383.4777.66
BaselineNFE / Gen Len=1/22026.0167.5577.6379.8375
EvoToken (10k FT)NFE / Gen Len=1/42026.0164.8275.7472.3370.96
FT-Base (10k FT)NFE / Gen Len=1/22026.0163.9178.627973.84
BaselineNFE / Gen Len=1/42026.0159.1468.2366.5764.65
FT-Base (10k FT)NFE / Gen Len=1/42026.0149.0562.1761.8757.7