In 18:58, I think there is a lot of cost if you passed entire URLs (images) to GPT vision for each QnA, we can tricky, once user upload an image, GPT Vision will extract entire information + image name (as ID) and save to table. How if the user ask a question, there 2 options for the response: 1. Retrieve entire all information from the table and send to GPT Text Generation 2. Use RAG concept, so for each image uploaded we can embbed the extracted information from GPT Vision
New Subscriber ! Thank You !!
Brilliant. It'd be cool to have something similar for extracting the line items from the receipt and saving the parsed data in a table.
In 18:58, I think there is a lot of cost if you passed entire URLs (images) to GPT vision for each QnA, we can tricky, once user upload an image, GPT Vision will extract entire information + image name (as ID) and save to table. How if the user ask a question, there 2 options for the response:
1. Retrieve entire all information from the table and send to GPT Text Generation
2. Use RAG concept, so for each image uploaded we can embbed the extracted information from GPT Vision
Excellent point, we have a set of RAG templates and videos coming very soon that will provide better alternatives for use cases such as this.
Please integrate with flutterflow also, otherwise it not help us!!