Federated Reasoning Distillation Framework with Model Learnability-Aware Data Allocation

TL;DR

This paper introduces LaDa, a federated reasoning distillation framework that adaptively allocates data based on model learnability gaps and employs domain adaptive distillation to enhance reasoning transfer between large and small language models.

Contribution

It proposes a novel learnability-aware data allocation method and a domain adaptive reasoning distillation technique for improved federated LLM and SLM collaboration.

Findings

Effective bidirectional knowledge transfer facilitated

Enhanced reasoning pattern acquisition in SLMs

Flexible adaptation to local data domains achieved

Abstract

Data allocation plays a critical role in federated large language model (LLM) and small language models (SLMs) reasoning collaboration. Nevertheless, existing data allocation methods fail to address an under-explored challenge in collaboration: bidirectional model learnability gap, where client-side SLMs cannot identify high-reward samples matching their learnability constraints for effective knowledge transfer from LLMs, while LLMs struggle to select samples contributing novel knowledge beyond their existing data. Furthermore, these collaboration frameworks face another key challenge: domain-agnostic reasoning transfer, where existing reasoning transfer methods fail to flexibly adapt to the local domain data, preventing SLMs from effectively acquiring step-by-step reasoning abilities within from general LLM. To address these challenges, we propose LaDa, a federated reasoning distillation framework with model learnability-aware data allocation. It introduces a model learnability-aware data filter that adaptively allocates high-reward samples based on the learnability gap between each SLM and LLM pair, effectively facilitating bidirectional knowledge transfer. We further design a domain adaptive reasoning distillation method that aligns joint probabilities of reasoning paths on filtered high-reward samples through contrastive distillation learning between SLM and LLM, enabling SLM to capture underlying reasoning patterns under local data distribution. LaDa operates as a plug-in module for existing collaboration frameworks, adapting knowledge transfer based on model learnability gaps.