Style Vectors for Steering Generative Large Language Model
This research addresses the need for more adaptive AI systems by enabling style control without complex training, though it is incremental as it builds on existing activation engineering methods.
The paper tackles the problem of steering large language model outputs towards specific styles like sentiment or writing style by adding style vectors to hidden layer activations, showing that these vectors can be computed simply from recorded activations and effectively influence generated text in a nuanced way.
This research explores strategies for steering the output of large language models (LLMs) towards specific styles, such as sentiment, emotion, or writing style, by adding style vectors to the activations of hidden layers during text generation. We show that style vectors can be simply computed from recorded layer activations for input texts in a specific style in contrast to more complex training-based approaches. Through a series of experiments, we demonstrate the effectiveness of activation engineering using such style vectors to influence the style of generated text in a nuanced and parameterisable way, distinguishing it from prompt engineering. The presented research constitutes a significant step towards developing more adaptive and effective AI-empowered interactive systems.