List your company

Mathematical Reasoning on GSM8K (cross)

81.5AccuracyBaseline
62.613667.516872.4277.3232Apr 29, 2026

Evaluation results

MethodLinks
BaselineN=16, setting=English occurrence as language confusion, aggregation=Averaged across four models and four target languages2026.0481.5
TLPON=16, setting=English occurrence as language confusion, aggregation=Averaged across four models and four target languages2026.0478.71
ORPON=16, setting=English occurrence as language confusion, aggregation=Averaged across four models and four target languages2026.0476.5
DPON=16, setting=English occurrence as language confusion, aggregation=Averaged across four models and four target languages2026.0474.44
SFTN=16, setting=English occurrence as language confusion, aggregation=Averaged across four models and four target languages2026.0463.34