This document covers the details of the experiment used in this repo.
- Step1: In the step1 of experiment we use product images with GPT4V (LMM) prompt to extract an in-depth, textual description of a product.
- Step2: In second step, image textual description, product attributes and keywords are used as input in GPT3.5 (LLM) prompt to get correct/incorrect keywords list as a response.