Instruction tuning is commonly assumed to endow language models with a domain-general ability to follow instructions, yet the underlying mechanism remains poorly understood. Does instruction-following rely on a universal mechanism or compositional skill deployment? We investigate this through diagnostic probing across nine diverse tasks in three instruction-tuned models.

Our analysis provides converging evidence against a universal mechanism. First, general probes trained across all tasks show selective rather than uniform deficits relative to task-specific specialists, indicating that representational sharing is partial and structured rather than global. Second, cross-task transfer is weak and clustered by skill similarity. Third, causal ablation reveals sparse asymmetric dependencies rather than shared representations. Tasks also stratify by complexity across layers, with structural constraints emerging early and semantic tasks emerging late. Finally, temporal analysis shows that the constraint signal becomes decodable only once generation is under way, and remains so throughout the response.

These findings indicate that instruction-following is better characterized as skillful coordination of diverse linguistic capabilities rather than deployment of a single abstract constraint-checking process.

In this paper we present and discuss the results achieved by the “Augustine of Hippo” team at SemEval-2023 Task 4 about human value detection. In particular, we provide a quantitative and qualitative reviews of the results obtained by SuperASKE, discussing respectively performance metrics and classification errors. Finally, we present our main contribution: an explainable and unsupervised approach mapping arguments to concepts, followed by a supervised classification model mapping concepts to human values.