分步教学提升多步检索模型推理能力,让小模型逼近大模型表现。
STEPER: Step-wise Knowledge Distillation for Enhancing Reasoning Ability in Multi-Step Retrieval-Augmented Language Models
- 按步骤分阶段监督,匹配不同阶段的信息需求
- 80亿参数模型性能媲美700亿参数教师模型
- 适合需要逐步推理的复杂问答系统
回答现实世界中的复杂问题需要分步检索并整合相关信息以生成有依据的回答。然而,现有知识蒸馏方法忽略了多步检索框架中各阶段对不同推理能力的需求,限制了知识迁移效果。为此,我们提出针对多步检索增强型语言模型的分步知识蒸馏方法(StepER)。StepER通过分步监督,使学生模型与各阶段不断变化的信息与推理需求对齐;同时引入难度感知训练,优先优化合适阶段的学习。该方法可适配多种多步检索增强模型,包括使用检索查询构建推理路径或拆解问题的模型。大量实验表明,StepER在多跳问答基准上优于已有方法,80亿参数模型表现接近700亿参数教师模型。
原文摘要 · Abstract (English)
Answering complex real-world questions requires step-by-step retrieval and integration of relevant information to generate well-grounded responses. However, existing knowledge distillation methods overlook the need for different reasoning abilities at different steps, hindering transfer in multi-step retrieval-augmented frameworks. To address this, we propose Stepwise Knowledge Distillation for Enhancing Reasoning Ability in Multi-Step Retrieval-Augmented Language Models (StepER). StepER employs step-wise supervision to align with evolving information and reasoning demands across stages. Additionally, it incorporates difficulty-aware training to progressively optimize learning by prioritizing suitable steps. Our method is adaptable to various multi-step retrieval-augmented language models, including those that use retrieval queries for reasoning paths or decomposed questions. Extensive experiments show that StepER outperforms prior methods on multi-hop QA benchmarks, with an 8B model achieving performance comparable to a 70B teacher model.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品;中文卡片由大模型生成,请以原文为准。