A specialist model laser-focused on relation extraction from text. It operates as a 'reader' component in a pipeline architecture, taking candidate entity spans and producing structured relation predictions. Built on DeBERTa v3 Large, it brings strong contextual understanding to the task, though it's designed to work within the ReLiK framework rather than as a standalone general-purpose tool.