merveenoyan / smol-vision

Recipes for shrinking, optimizing, customizing cutting edge vision models. 💜
Apache License 2.0
912 stars 88 forks source link

Querying Multiple PDFs Simultaneously with RAG Implementation in ColPali_+_Qwen2_VL.ipynb #16

Closed jaswanth2003 closed 2 months ago

jaswanth2003 commented 2 months ago

Thank you for providing the ColPali_+_Qwen2_VL.ipynb notebook. It has been incredibly helpful in understanding and applying Retrieval Augmented Generation.

I am wondering if it is possible to extend the current implementation to ask one question from multiple PDFs at the same time. For instance, retrieving information from both pdf1 and pdf2 at the same time . Currently, the notebooks appear to handle retrieval from a single document or a specific context window, even if there are multiple PDFs in the database. However, in many real-world scenarios, it would be beneficial to retrieve information from multiple documents simultaneously.

merveenoyan commented 2 months ago

@jaswanth2003 I don't understand the motivation or lack of limitations here, in this tutorial you are asking questions to 10 pages (from what I remember of the PDF), retrieve one page and answer question.

jaswanth2003 commented 2 months ago

Yes, got it! I was confused with retrieving multiple pages and feeding them. Now I understand that I can feed multiple pages at the same time.