Hi, I've read your paper and found it very interesting. I'm particularly interested in how you obtained the textual description rationales (td_rationale) and commonsense rationales (cs_rationale) from GPT-3.5.
While I see the Few-Shot and Zero-Shot prompting templates in Table 8, could you please share the specific prompts you used to generate these rationales in your provided dataset?
Thank you for making this research publicly available!