Hey everyone! I am super excited to share a new paper summary video of "Lost in the Middle: How Language Models use Long Contexts" - Liu et al. 2023! This paper explores the impact of Search quality on Large Language Models in RAG (Retrieval-Augmented Generation)!
The authors find this amazing U-shaped performance where the language model performs very poorly when the relevant information is in the middle of the context (e.g. search result 10 out of 20), but performs extremely well when the information is either 1st, or oddly at the very end of the context!
This video explores the experimental details of the paper as well as an explanation of new Weaviate features such as AutoCut, Re-Rankers, and Hybrid Rank Fusion to help you get better search results and avoid getting stuck in the middle when building RAG applications!
I hope you enjoy the video! More than happy to answer any questions or discuss any ideas you have about this!
https://youtu.be/Kf3LeaUGwlg