Prompt Management

LLMLingua uses a compact language model to remove unnecessary tokens in prompts, leading to efficient inference with large language models and up to 20x compression without significant performance loss.