Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

World Knowledge and Reading Comprehension on NQ, MMLU STEM, ARC, SciQ, LogiQA, BoolQ

29.81NQ Accuracy

Mistral-7B Base

12.608417.074221.5426.0058Feb 12, 2024
Updated 1mo ago

Evaluation Results

MethodLinks
2024.02
29.8152.3984.6857.2597.430.2683.5854.3
2024.02
29.2252.6284.7257.2597.330.2673.7653.54
2024.02
29.1752.1784.1856.6697.231.0383.8254.91
2024.02
29.0652.384.1855.296.831.0383.1255.19
2024.02
28.8952.0185.4857.7697.331.1882.8154.03
2024.02
26.0740.0980.7750.0996.727.6578.4146.62
2024.02
26.0137.0880.7249.7496.826.5777.6844.99
2024.02
25.9637.8480.4350.696.827.6577.7144.97
2024.02
25.8440.6680.0949.7496.727.9677.1947.07
2024.02
25.7636.6380.9848.9896.526.7372.5444.27
2024.02
14.8836.677.6146.596.325.3569.5442.92
2024.02
14.3233.8177.9546.7696.224.4268.5341.67
2024.02
13.836.5477.445.3996.426.2768.3542.95
2024.02
13.2734.4477.2745.8295.923.554.9239.71
2024.02
13.2736.0976.8146.0896.127.1971.2843.16