用轻量适配器训练模型自解释，无需修改原模型
Learning Self-Interpretation from Interpretability Artifacts: Training Lightweight Adapters on Vector-Label Pairs
- 在冻结语言模型前提下，用可解释性特征向量训练轻量适配器
- 适配器生成的特征标签准确率达70%，主题识别召回率94%
- 适用于不同模型规模，适合研究模型内部推理过程
自解释方法促使语言模型描述自身内部状态，但因超参数敏感而不可靠。我们发现，在保持语言模型完全冻结的前提下，仅需训练轻量适配器即可实现跨任务、跨模型家族的可靠自解释。一个仅含 $d_ ext{model}+1$ 参数的标量仿射适配器即足够：训练后的适配器生成的稀疏自编码器特征标签，其生成评分达到70%（70B规模），优于训练标签本身的50%；主题识别召回率@1达94%，远超未训练基线的1%；能解码多跳推理中既不在提示也不在响应中的桥接实体，揭示隐式推理路径而无需链式思维。所学偏置向量单独贡献了85%的性能提升，且更简单的适配器泛化能力优于更复杂版本。通过提示描述控制模型知识后，发现自解释性能提升超过7B至72B参数规模带来的能力增长。结果表明，自解释能力随规模提升，且无需修改被解释模型。
原文摘要 · Abstract (English)
Self-interpretation methods prompt language models to describe their own internal states, but remain unreliable due to hyperparameter sensitivity. We show that training lightweight adapters on interpretability artifacts, while keeping the LM entirely frozen, yields reliable self-interpretation across tasks and model families. A scalar affine adapter with just $d_\text{model}+1$ parameters suffices: trained adapters generate sparse autoencoder feature labels that outperform the training labels themselves (70% vs 50% generation scoring at 70B scale), identify topics with 94% recall@1 versus 1% for untrained baselines, and decode bridge entities in multi-hop reasoning that appear in neither prompt nor response, surfacing implicit reasoning without chain-of-thought. The learned bias vector alone accounts for 85% of improvement, and simpler adapters generalize better than more expressive alternatives. Controlling for model knowledge via prompted descriptions, we find self-interpretation gains outpace capability gains from 7B to 72B parameters. Our results demonstrate that self-interpretation improves with scale, without modifying the model being interpreted.
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。