I am encountering an issue with OpenAI's document-based response system using openai assistant API. Despite uploading a file containing information on 20 hotels (totaling 15,000 words), when I ask the assistant to provide me with the names of all the hotels, it only returns the names of the first two hotels. I seek guidance on how to ensure that the entire file is considered in generating responses, rather than restricting output based on a single vector.
How can I ensure that the entire content of a document is considered instead of relying on a specific vector?
For example, if I have a PDF containing data on 20 hotels spanning 15,000 words, and I ask for a list of all the hotels, the response typically includes only two hotel names due to the limitation of considering only one vector at a time. I am seeking guidance on how to enable the system to process the entire document, ensuring that the response includes all 20 hotel names.