I have been getting “Error count = 5” in the decide_task_nature function’s unit test, but I do not understand what is wrong with the prompt. The outputs I’m getting for the demonstration queries after the unit test are shown below. I have no idea how “based” and “I” are being considered as labels. How do I debug this?
Query: Give me two sneakers with vibrant colors. Label Predicted: Based. Correct Label: technical Total Tokens: 153
Query: What are the most expensive clothes you have in your catalogue? Label Predicted: Based. Correct Label: technical Total Tokens: 157
Query: I have a green Dress and I like a suggestion on an accessory to match with it. Label Predicted: I. Correct Label: creative Total Tokens: 163
Query: Give me three trousers with vibrant colors you have in your catalogue. Label Predicted: Based. Correct Label: technical Total Tokens: 158
Query: Create a look for a woman walking in a park on a sunny day. It must be fresh due to hot weather. Label Predicted: I. Correct Label: creative Total Tokens: 169
