解决大模型稀疏自编码器训练中的特征吸收问题
Time-Aware Feature Selection: Adaptive Temporal Masking for Stable Sparse Autoencoder Training
- 动态追踪激活强度、频率和重建贡献，生成随时间演化的特征重要性分数
- 实验显示在Gemma-2-2b上吸收率显著低于TopK和JumpReLU方法
- 适合需要稳定可解释特征的模型可靠性分析场景
理解大语言模型的内部表征对确保其可靠性与安全性至关重要，稀疏自编码器（SAEs）作为有前景的可解释性方法正受到关注。然而，现有SAE训练方法存在特征吸收问题，即特征（或神经元）相互吸收以最小化$L_1$惩罚，导致难以一致识别和分析模型行为。本文提出自适应时间掩码（ATM），通过追踪激活幅度、频率和重建贡献来计算随时间演化的特征重要性分数，并基于统计阈值实施概率掩码机制，实现更自然的特征选择。在Gemma-2-2b模型上的大量实验表明，ATM在保持优异重构质量的同时，吸收得分显著低于TopK和JumpReLU SAE等现有方法。该结果确立了ATM作为学习神经网络中稳定、可解释特征的原理性解决方案，为更可靠的模型分析提供了基础。
原文摘要 · Abstract (English)
Understanding the internal representations of large language models is crucial for ensuring their reliability and safety, with sparse autoencoders (SAEs) emerging as a promising interpretability approach. However, current SAE training methods face feature absorption, where features (or neurons) are absorbed into each other to minimize $L_1$ penalty, making it difficult to consistently identify and analyze model behaviors. We introduce Adaptive Temporal Masking (ATM), a novel training approach that dynamically adjusts feature selection by tracking activation magnitudes, frequencies, and reconstruction contributions to compute importance scores that evolve over time. ATM applies a probabilistic masking mechanism based on statistical thresholding of these importance scores, creating a more natural feature selection process. Through extensive experiments on the Gemma-2-2b model, we demonstrate that ATM achieves substantially lower absorption scores compared to existing methods like TopK and JumpReLU SAEs, while maintaining excellent reconstruction quality. These results establish ATM as a principled solution for learning stable, interpretable features in neural networks, providing a foundation for more reliable model analysis.
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。