Logical Reasoning On Big Bench Logical

Accuracy

Results

Performance results of various models on this benchmark

Comparison Table

Model Name	Accuracy
training-compute-optimal-large-language	72.1
scaling-language-models-methods-analysis-1	58.9