MAGDi: Structured Distillation of Multi-Agent Interaction Graphs Improves Reasoning in Smaller Language Models

Justin Chen, Swarnadeep Saha, Elias Stengel-Eskin, Mohit Bansal
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:7220-7235, 2024.

Abstract

Multi-agent interactions between Large Language Model (LLM) agents have shown major improvements on diverse reasoning tasks. However, these involve long generations from multiple models across several rounds, making them expensive. Moreover, these multi-agent approaches fail to provide a final, single model for efficient inference. To address this, we introduce MAGDi, a new method for structured distillation of the reasoning interactions between multiple LLMs into smaller LMs. MAGDi teaches smaller models by representing multi-agent interactions as graphs, augmenting a base student model with a graph encoder, and distilling knowledge using three objective functions: next-token prediction, a contrastive loss between correct and incorrect reasoning, and a graph-based objective to model the interaction structure. Experiments on seven widely used commonsense and math reasoning benchmarks show that MAGDi improves the reasoning capabilities of smaller models, outperforming several methods that distill from a single teacher and multiple teachers. Moreover, MAGDi also demonstrates an order of magnitude higher efficiency over its teachers. We conduct extensive analyses to show that MAGDi (1) enhances the generalizability to out-of-domain tasks, (2) scales positively with the size and strength of the base student model, and (3) obtains larger improvements (via our multi-teacher training) when applying self-consistency – an inference technique that relies on model diversity.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-chen24ah, title = {{MAGD}i: Structured Distillation of Multi-Agent Interaction Graphs Improves Reasoning in Smaller Language Models}, author = {Chen, Justin and Saha, Swarnadeep and Stengel-Eskin, Elias and Bansal, Mohit}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {7220--7235}, year = {2024}, editor = {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine and Weller, Adrian and Oliver, Nuria and Scarlett, Jonathan and Berkenkamp, Felix}, volume = {235}, series = {Proceedings of Machine Learning Research}, month = {21--27 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v235/main/assets/chen24ah/chen24ah.pdf}, url = {https://proceedings.mlr.press/v235/chen24ah.html}, abstract = {Multi-agent interactions between Large Language Model (LLM) agents have shown major improvements on diverse reasoning tasks. However, these involve long generations from multiple models across several rounds, making them expensive. Moreover, these multi-agent approaches fail to provide a final, single model for efficient inference. To address this, we introduce MAGDi, a new method for structured distillation of the reasoning interactions between multiple LLMs into smaller LMs. MAGDi teaches smaller models by representing multi-agent interactions as graphs, augmenting a base student model with a graph encoder, and distilling knowledge using three objective functions: next-token prediction, a contrastive loss between correct and incorrect reasoning, and a graph-based objective to model the interaction structure. Experiments on seven widely used commonsense and math reasoning benchmarks show that MAGDi improves the reasoning capabilities of smaller models, outperforming several methods that distill from a single teacher and multiple teachers. Moreover, MAGDi also demonstrates an order of magnitude higher efficiency over its teachers. We conduct extensive analyses to show that MAGDi (1) enhances the generalizability to out-of-domain tasks, (2) scales positively with the size and strength of the base student model, and (3) obtains larger improvements (via our multi-teacher training) when applying self-consistency – an inference technique that relies on model diversity.} }
Endnote
%0 Conference Paper %T MAGDi: Structured Distillation of Multi-Agent Interaction Graphs Improves Reasoning in Smaller Language Models %A Justin Chen %A Swarnadeep Saha %A Elias Stengel-Eskin %A Mohit Bansal %B Proceedings of the 41st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2024 %E Ruslan Salakhutdinov %E Zico Kolter %E Katherine Heller %E Adrian Weller %E Nuria Oliver %E Jonathan Scarlett %E Felix Berkenkamp %F pmlr-v235-chen24ah %I PMLR %P 7220--7235 %U https://proceedings.mlr.press/v235/chen24ah.html %V 235 %X Multi-agent interactions between Large Language Model (LLM) agents have shown major improvements on diverse reasoning tasks. However, these involve long generations from multiple models across several rounds, making them expensive. Moreover, these multi-agent approaches fail to provide a final, single model for efficient inference. To address this, we introduce MAGDi, a new method for structured distillation of the reasoning interactions between multiple LLMs into smaller LMs. MAGDi teaches smaller models by representing multi-agent interactions as graphs, augmenting a base student model with a graph encoder, and distilling knowledge using three objective functions: next-token prediction, a contrastive loss between correct and incorrect reasoning, and a graph-based objective to model the interaction structure. Experiments on seven widely used commonsense and math reasoning benchmarks show that MAGDi improves the reasoning capabilities of smaller models, outperforming several methods that distill from a single teacher and multiple teachers. Moreover, MAGDi also demonstrates an order of magnitude higher efficiency over its teachers. We conduct extensive analyses to show that MAGDi (1) enhances the generalizability to out-of-domain tasks, (2) scales positively with the size and strength of the base student model, and (3) obtains larger improvements (via our multi-teacher training) when applying self-consistency – an inference technique that relies on model diversity.
APA
Chen, J., Saha, S., Stengel-Eskin, E. & Bansal, M.. (2024). MAGDi: Structured Distillation of Multi-Agent Interaction Graphs Improves Reasoning in Smaller Language Models. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:7220-7235 Available from https://proceedings.mlr.press/v235/chen24ah.html.

Related Material