BTR: Binary Token Representations for Efficient Retrieval Augmented Language Models
This addresses efficiency and scalability issues for users of retrieval-augmented language models, though it is incremental as it builds on existing retrieval augmentation methods.
The paper tackles the problem of slow and unscalable retrieval-augmented language models by introducing binary token representations (BTR), which accelerate state-of-the-art inference by up to 4x and reduce storage by over 100x while maintaining over 95% task performance on five knowledge-intensive NLP tasks.
Retrieval augmentation addresses many critical problems in large language models such as hallucination, staleness, and privacy leaks. However, running retrieval-augmented language models (LMs) is slow and difficult to scale due to processing large amounts of retrieved text. We introduce binary token representations (BTR), which use 1-bit vectors to precompute every token in passages, significantly reducing computation during inference. Despite the potential loss of accuracy, our new calibration techniques and training objectives restore performance. Combined with offline and runtime compression, this only requires 127GB of disk space for encoding 3 billion tokens in Wikipedia. Our experiments show that on five knowledge-intensive NLP tasks, BTR accelerates state-of-the-art inference by up to 4x and reduces storage by over 100x while maintaining over 95% task performance.