The Geometry of Prompting: Unveiling Distinct Mechanisms of Task Adaptation in Language Models
This work addresses the problem of understanding how language models adapt to different tasks through prompting, which is significant for researchers and developers of large language models.
This study investigates the internal mechanism of task adaptation in language models through prompting, revealing distinct representational mechanisms for different prompting techniques, with a focus on input distribution samples and label semantics. The analysis shows synergistic and interfering interactions between tasks on the representational level.
Decoder-only language models have the ability to dynamically switch between various computational tasks based on input prompts. Despite many successful applications of prompting, there is very limited understanding of the internal mechanism behind such flexibility. In this work, we investigate how different prompting methods affect the geometry of representations in these models. Employing a framework grounded in statistical physics, we reveal that various prompting techniques, while achieving similar performance, operate through distinct representational mechanisms for task adaptation. Our analysis highlights the critical role of input distribution samples and label semantics in few-shot in-context learning. We also demonstrate evidence of synergistic and interfering interactions between different tasks on the representational level. Our work contributes to the theoretical understanding of large language models and lays the groundwork for developing more effective, representation-aware prompting strategies.