发现稀疏自编码器中特征死亡主因是激活异常值，可被均值中心化解决。
On the Relationship Between Activation Outliers and Feature Death in Sparse Autoencoders
- 通过激活均值与方差比γ量化异常值严重度，预测特征死亡率。
- γ值高时超过70%特征无法激活，且训练难以恢复。
- 训练前减去激活均值可彻底避免特征死亡，适合所有模型。
稀疏自编码器（SAEs）将神经网络激活分解为可解释特征，但许多特征从不激活，即特征死亡问题，浪费字典容量并可能重引入超叠加。不同模型间死亡率差异巨大：GPT-2接近零，AlphaFold3超过70%，即使配置相同。我们发现，维度级激活异常值（其均值幅度远大于单个激活的波动）导致该现象，因初始化时各特征对激活均值的对齐程度不同，反向对齐的特征获得永久负预激活而永不触发。我们定义异常值严重度为γ = ||μ||/||σ||；γ能有效预测初始死亡率（在454个模型层组合上，基于TopK的死亡率相关性Spearman ρ=0.89，基于ReLU的ρ=0.82）。死掉的特征可在训练中恢复，但需自编码器偏置学习激活均值，当γ较高时过程极慢。均值中心化（减去激活均值）可绕过此问题，在所有测试模型中完全消除异常值引起的死亡，验证了机制并为该预处理步骤提供了理论依据。
原文摘要 · Abstract (English)
Sparse autoencoders (SAEs) decompose neural network activations into interpretable features, but many learned features never activate, a problem called feature death that wastes dictionary capacity and can reintroduce superposition. Death rates vary dramatically between models: near-zero on GPT-2, over 70% on AlphaFold3 with identical configurations. We find that dimension-level activation outliers (dimensions whose mean magnitude is large relative to per-token variation) cause this by shifting pre-activations at initialization based on each feature's alignment with the activation mean. Features anti-aligned with the mean receive permanently negative pre-activations and never fire. We formalize outlier severity as $γ= \|μ\|/\|σ\|$; it predicts initial death rates (Spearman $ρ= 0.89$ for dead-by-TopK, $0.82$ for dead-by-ReLU) across 454 model-layer combinations spanning language, vision, protein, and genomic models. Dead features can revive during training, but recovery requires the SAE bias to learn the activation mean, a process that is prohibitively slow at high $γ$. Mean-centering (subtracting the activation mean) sidesteps this and eliminates outlier-induced death across all tested models, confirming the mechanism and providing a principled basis for when and why this preprocessing step is necessary.
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。