MAMM-Refine: A Recipe for Improving Faithfulness in Generation with Multi-Agent Collaboration
This addresses the problem of factual inconsistencies in generated text for users of summarization and QA systems, presenting an incremental improvement through multi-agent collaboration.
The paper tackled improving faithfulness in long-form generation tasks like summarization and question-answering by extending multi-agent collaboration to refinement, resulting in a recipe called MAMM-Refine that significantly boosted performance on three summarization datasets and long-form question answering.
Multi-agent collaboration among models has shown promise in reasoning tasks but is underexplored in long-form generation tasks like summarization and question-answering. We extend multi-agent multi-model reasoning to generation, specifically to improving faithfulness through refinement, i.e., revising model-generated outputs to remove factual inconsistencies. We investigate how iterative collaboration among multiple instances and types of large language models (LLMs) enhances subtasks in the refinement process, such as error detection, critiquing unfaithful sentences, and making corrections based on critiques. We design intrinsic evaluations for each subtask, with our findings indicating that both multi-agent (multiple instances) and multi-model (diverse LLM types) approaches benefit error detection and critiquing. Additionally, reframing critiquing and refinement as reranking rather than generation tasks improves multi-agent performance. We consolidate these insights into a final "recipe" called Multi-Agent Multi-Model Refinement (MAMM-Refine), where multi-agent and multi-model collaboration significantly boosts performance on three summarization datasets as well as on long-form question answering, demonstrating the effectiveness and generalizability of our recipe.