List your company

Mathematical Reasoning on MultiArith (Punctuation Robustness)

99.44Accuracy (Clean)Llama3.1
34.731251.530668.3385.1294Jan 14, 2025

Evaluation results

MethodLinks
Llama3.12025.0199.4494.4491.6683.889.67
Qwen2.52025.0197.2294.4485.5583.8811.04
Mathstral2025.0196.1192.7786.1187.229.47
Llama32025.019592.7791.6688.337.79
DeepSeek2025.0193.889092.7788.888.28
Gemma22025.0189.4482.7778.8872.2219.45
Mistral2025.0173.8872.7771.1165.5523.66
Zephyr2025.0137.2222.2216.1112.7777.1