Request Correction

Augmenting LLM Reasoning with Dynamic Notes Writing for Complex MultiHop QA

Iterative RAG for multi-hop question answering faces challenges with lengthy contexts and the buildup of irrelevant information. This hinders a model’s capacity to process and reason over retrieved content and limits performance. While recent methods focus on compressing retrieved information, they are either restricted to single-round RAG, require finetuning or lack scalability in iterative RAG. To address these, we propose NotesWriting, a method that generates concise and relevant notes from retrieved documents at each step, thereby reducing noise and retaining only essential information. This increases the effective context length of Large Language Models (LLMs), allowing them to reason and plan more effectively while processing larger volumes of input text due to the compression in the form of notes. NotesWriting is framework agnostic and can be integrated with different iterative RAG methods. We demonstrate its effectiveness with three iterative RAG methods, across two models and four evaluation datasets. NotesWriting yields an average improvement of 15.6 percentage points overall, by scaling the amount of ingested information.