prompt tuning

What is prompt tuning?

Prompt tuning is a technique used in natural language processing (NLP) to adapt pre-trained language models (PLMs) to specific downstream tasks. Instead of fine-tuning all the parameters of a large PLM, prompt tuning involves learning a small, task-specific 'prompt' (or prefix) that is prepended to the input text. The PLM's parameters remain frozen, and only the prompt is optimized during training. This approach significantly reduces the computational cost and storage requirements compared to full fine-tuning, while still achieving competitive performance. It's commonly used in scenarios where computational resources are limited, or when adapting a model to many different tasks is desired. The prompt guides the PLM to generate the desired output for the given task.