Search papers, labs, and topics across Lattice.
This paper investigates how hierarchical relations are encoded in language models by training linear transformations specific to hierarchical depth and semantic domain. They extend prior work by analyzing multi-token entities and cross-layer representations, evaluating both in-domain generalization and cross-domain transfer. The key finding is that while hierarchical information is encoded in a low-dimensional, domain-specific subspace, the representation of hierarchy is highly similar across these subspaces, suggesting a general encoding mechanism.
Despite domain-specific encoding subspaces, language models represent hierarchical relationships with surprisingly consistent linear transformations across different semantic domains.
We investigate how and to what extent hierarchical relations (e.g., Japan $\subset$ Eastern Asia $\subset$ Asia) are encoded in the internal representations of language models. Building on Linear Relational Concepts, we train linear transformations specific to each hierarchical depth and semantic domain, and characterize representational differences associated with hierarchical relations by comparing these transformations. Going beyond prior work on the representational geometry of hierarchies in LMs, our analysis covers multi-token entities and cross-layer representations. Across multiple domains we learn such transformations and evaluate in-domain generalization to unseen data and cross-domain transfer. Experiments show that, within a domain, hierarchical relations can be linearly recovered from model representations. We then analyze how hierarchical information is encoded in representation space. We find that it is encoded in a relatively low-dimensional subspace and that this subspace tends to be domain-specific. Our main result is that hierarchy representation is highly similar across these domain-specific subspaces. Overall, we find that all models considered in our experiments encode concept hierarchies in the form of highly interpretable linear representations.
