Depth-Wise Emergence of Prediction-Centric Geometry in Large Language Models
This provides a mechanistic-geometric account of LLM dynamics, which is incremental for researchers in interpretability and model understanding.
The paper tackled the problem of understanding how large language models transform context into predictions by showing a depth-wise transition from context-processing to prediction-forming phases, accompanied by reorganization of representational geometry, with late-layer representations implementing a structured geometric code that enables selective causal control over token prediction.
We show that decoder-only large language models exhibit a depth-wise transition from context-processing to prediction-forming phases of computation accompanied by a reorganization of representational geometry. Using a unified framework combining geometric analysis with mechanistic intervention, we demonstrate that late-layer representations implement a structured geometric code that enables selective causal control over token prediction. Specifically, angular organization of the representation geometry parametrizes prediction distributional similarity, while representation norms encode context-specific information that does not determine prediction. Together, these results provide a mechanistic-geometric account of the dynamics of transforming context into predictions in LLMs.