通过迭代蒸馏提取稳定可用的稀疏特征，提升模型解释性。
Attribution-Guided Distillation of Matryoshka Sparse Autoencoders
- 用梯度归因筛选核心特征，仅保留对预测贡献大的最小集合。
- 7轮蒸馏后得到197个重复选中的稳定特征，性能优于基线。
- 适合需要可复用、可解释特征的模型分析与部署场景。
稀疏自编码器（SAEs）旨在将模型激活分解为单一语义、人类可读的特征。然而，实际学习到的特征常存在冗余且在不同训练轮次和稀疏度下变化大，导致解释难以迁移与复用。本文提出蒸馏式套娃稀疏自编码器（DMSAE），通过迭代蒸馏循环：训练具有共享核心的套娃SAE，利用梯度X激活衡量每个特征在最内层重构中对下一步词损失的贡献，仅保留能解释固定比例归因的最小特征子集。每轮仅传递核心编码器权重，核心解码器及所有非核心隐变量重置。在Gemma-2-2B第12层残差流上，7轮蒸馏（500M tokens，65k宽度）生成了197个反复被选中的核心特征。使用该核心训练新SAE，在多个SAEBench指标上表现更优，证明一致的潜在特征集可在不同稀疏度间迁移。
原文摘要 · Abstract (English)
Sparse autoencoders (SAEs) aim to disentangle model activations into monosemantic, human-interpretable features. In practice, learned features are often redundant and vary across training runs and sparsity levels, which makes interpretations difficult to transfer and reuse. We introduce Distilled Matryoshka Sparse Autoencoders (DMSAEs), a training pipeline that distills a compact core of consistently useful features and reuses it to train new SAEs. DMSAEs run an iterative distillation cycle: train a Matryoshka SAE with a shared core, use gradient X activation to measure each feature's contribution to next-token loss in the most nested reconstruction, and keep only the smallest subset that explains a fixed fraction of the attribution. Only the core encoder weight vectors are transferred across cycles; the core decoder and all non-core latents are reinitialized each time. On Gemma-2-2B layer 12 residual stream activations, seven cycles of distillation (500M tokens, 65k width) yielded a distilled core of 197 features that were repeatedly selected. Training using this distilled core improves several SAEBench metrics and demonstrates that consistent sets of latent features can be transferred across sparsity levels
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。