CRUXEval-Input-CoT

CRUXEval-Input-CoT is a code reasoning benchmark that evaluates Large Language Models' ability to predict function inputs using chain-of-thought reasoning. This variant of CRUXEval focuses on input prediction tasks where models must determine what inputs would produce given outputs through systematic reasoning. The benchmark tests reverse engineering capabilities and logical deduction skills in code understanding scenarios.