I’m having multiple markdown files that contains information valuable as context for the assistant. E.g. one file contains a scraped website another has a FAQ list with answers.
I’m wondering if I should upload those as separate files or should I merge them to one. I can imagine that having them separate might result in a longer response time because the model may try to look into both files. On the other hand a separation could tell the model where to look for which information to become more precise.
The later is actually a second question I’m wondering. Is it possible (i.e… should we) refer to the attached files in the instructions like “In the file faq.md you find common questions and answers and the file website.md might be helpful to you to answer questions around the website.”? Are the file names even “transparent” to the model or should I replace them with the internal file IDs? Or does this make no sense anyway to hint about the files in the instructions?