Universal Activation Verbalizer: A Unified Framework for Cross-Model Activation Explanation
This work addresses the limitation of self-explanation in activation verbalization, enabling cross-model explanation for diverse model families and scales.
UAV introduces a shared decoder framework that explains activations from heterogeneous models via a lightweight adapter, achieving competitive performance with self-explanation baselines across classification, fact retrieval, and gist summarization while enabling cross-model verbalization.
Activation verbalization explains hidden representations in natural language, but existing methods are mostly limited to self-explanation, where each model explains only its own activations. We introduce Universal Activation Verbalizer (UAV), a framework that uses a shared decoder to explain activations from heterogeneous donor models. UAV learns a lightweight adapter that converts donor activations into soft tokens in decoder's embedding space, and further supports adapter-only transfer by reusing a frozen decoder-side LoRA while training only a new adapter for another donor. Across classification, fact retrieval, and gist summarization, UAV remains competitive with strong self-explanation baselines while enabling cross-model verbalization across model families and scales. Ablations show that decoder-side tuning mainly improves task behavior, whereas the adapter provides the activation-grounded factual and semantic information needed for faithful explanations.