Figure 3

Bar graph showing the success rates of various prompting strategies ('Direct Prompt', 'CoT', and 'Modified CoT') for different 'calculation' question subtypes. Each bar represents the average success rate of a particular prompting strategy for a specific subtype.