小幅度参数调整可激发大模型推理能力，且信号可解释。
Rank-1 LoRAs Encode Interpretable Reasoning Signals
- 用单秩LoRA仅调整少量参数，实现高效推理能力提升。
- 在Qwen-2.5-32B上恢复73-90%的推理基准性能。
- 发现激活信号与推理行为直接相关，适合研究模型机理。
推理模型通过推理时计算显著提升语言模型在复杂逻辑任务上的表现，已成为前沿大模型的主流范式。然而，其性能提升的机制尚不明确。本文表明，多数推理能力可通过对基础模型参数进行微小的单秩调整来实现，且这些调整具有可解释性。具体地，我们使用单秩LoRA为Qwen-2.5-32B-Instruct构建极简参数适配器，在推理基准测试中恢复了73%-90%的全参数微调性能。我们发现该LoRA的激活模式与MLP神经元类似，能精准响应特定推理行为。进一步训练稀疏自编码器分析其完整激活状态，识别出细粒度且单一语义的特征。结果表明，推理能力主要源于基础参数的微小变化，且这些变化影响关键行为。本工作揭示，参数高效训练方法可作为探查语言模型行为与动态的有力工具。
原文摘要 · Abstract (English)
Reasoning models leverage inference-time compute to significantly enhance the performance of language models on difficult logical tasks, and have become a dominating paradigm in frontier LLMs. Despite their wide adoption, the mechanisms underpinning the enhanced performance of these reasoning models are not well understood. In this work, we show that the majority of new capabilities in reasoning models can be elicited by small, single-rank changes to base model parameters, with many of these changes being interpretable. Specifically, we use a rank-1 LoRA to create a minimal parameter adapter for Qwen-2.5-32B-Instruct which recovers 73-90% of reasoning-benchmark performance compared to a full parameter finetune. We find that the activations of this LoRA are as interpretable as MLP neurons, and fire for reasoning-specific behaviors. Finally, we train a sparse autoencoder on the entire activation state of this LoRA and identify fine-grained and monosemantic features. Our findings highlight that reasoning performance can arise largely from minimal changes to base model parameters, and explore what these changes affect. More broadly, our work shows that parameter-efficient training methods can be used as a targeted lens for uncovering fundamental insights about language model behavior and dynamics.
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。