ICL Markup: Structuring In-Context Learning using Soft-Token Tags
This work addresses the usability and efficiency challenges in in-context learning for users of large language models, though it is incremental as it builds on existing meta-learning and fine-tuning techniques.
The paper tackles the problem of overwhelming and arbitrary choices in in-context learning for large language models by introducing a method using soft-token tags to structure prompt templates, resulting in improved performance on tasks like few-shot intent detection and text classification in enterprise applications.
Large pretrained language models (LLMs) can be rapidly adapted to a wide variety of tasks via a text-to-text approach, where the instruction and input are fed to the model in natural language. Combined with in-context learning (ICL), this paradigm is impressively flexible and powerful. However, it also burdens users with an overwhelming number of choices, many of them arbitrary. Inspired by markup languages like HTML, we contribute a method of using soft-token tags to compose prompt templates. This approach reduces arbitrary decisions and streamlines the application of ICL. Our method is a form of meta-learning for ICL; it learns these tags in advance during a parameter-efficient fine-tuning ``warm-up'' process. The tags can subsequently be used in templates for ICL on new, unseen tasks without any additional fine-tuning. Our experiments with this approach yield promising initial results, improving LLM performance on important enterprise applications such as few-shot and open-world intent detection, as well as text classification in news and legal domains.