SMAGDi: Socratic Multi Agent Interaction Graph Distillation for Efficient High Accuracy Reasoning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Aluru, Aayush, Malik, Myra, Patankar, Samarth, Kim, Spencer, Zhu, Kevin, O'Brien, Sean, Sharma, Vasu
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914143320145920
author Aluru, Aayush
Malik, Myra
Patankar, Samarth
Kim, Spencer
Zhu, Kevin
O'Brien, Sean
Sharma, Vasu
author_facet Aluru, Aayush
Malik, Myra
Patankar, Samarth
Kim, Spencer
Zhu, Kevin
O'Brien, Sean
Sharma, Vasu
contents Multi-agent systems (MAS) often achieve higher reasoning accuracy than single models, but their reliance on repeated debates across agents makes them computationally expensive. We introduce SMAGDi, a distillation framework that transfers the debate dynamics of a five-agent Llama-based MAS into a compact Socratic decomposer-solver student. SMAGDi represents debate traces as directed interaction graphs, where nodes encode intermediate reasoning steps with correctness labels and edges capture continuity and cross-agent influence. The student is trained with a composite objective combining language modeling, graph-based supervision, contrastive reasoning, and embedding alignment to preserve both fluency and structured reasoning. On StrategyQA and MMLU, SMAGDi compresses a 40B multi-agent system into a 6B student while retaining 88% of its accuracy, substantially outperforming prior distillation methods such as MAGDi, standard KD, and fine-tuned baselines. These results highlight that explicitly modeling interaction graphs and Socratic decomposition enable small models to inherit the accuracy benefits of multi-agent debate while remaining efficient enough for real-world deployment.
format Preprint
id arxiv_https___arxiv_org_abs_2511_05528
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SMAGDi: Socratic Multi Agent Interaction Graph Distillation for Efficient High Accuracy Reasoning
Aluru, Aayush
Malik, Myra
Patankar, Samarth
Kim, Spencer
Zhu, Kevin
O'Brien, Sean
Sharma, Vasu
Artificial Intelligence
Multi-agent systems (MAS) often achieve higher reasoning accuracy than single models, but their reliance on repeated debates across agents makes them computationally expensive. We introduce SMAGDi, a distillation framework that transfers the debate dynamics of a five-agent Llama-based MAS into a compact Socratic decomposer-solver student. SMAGDi represents debate traces as directed interaction graphs, where nodes encode intermediate reasoning steps with correctness labels and edges capture continuity and cross-agent influence. The student is trained with a composite objective combining language modeling, graph-based supervision, contrastive reasoning, and embedding alignment to preserve both fluency and structured reasoning. On StrategyQA and MMLU, SMAGDi compresses a 40B multi-agent system into a 6B student while retaining 88% of its accuracy, substantially outperforming prior distillation methods such as MAGDi, standard KD, and fine-tuned baselines. These results highlight that explicitly modeling interaction graphs and Socratic decomposition enable small models to inherit the accuracy benefits of multi-agent debate while remaining efficient enough for real-world deployment.
title SMAGDi: Socratic Multi Agent Interaction Graph Distillation for Efficient High Accuracy Reasoning
topic Artificial Intelligence
url https://arxiv.org/abs/2511.05528