Efficient Knowledge Feeding to Language Models: A Novel Integrated Encoder-Decoder Architecture
This addresses limitations in Retrieval-Augmented Generation models for handling extensive datasets, offering a more robust solution for AI applications requiring efficient knowledge integration.
The paper tackles the problem of efficiently feeding knowledge to language models by introducing an integrated encoder-decoder architecture with in-context vectors (ICV), which outperforms standard in-context learning and fine-tuning across tasks like question-answering while reducing computational costs and memory requirements.
This paper introduces a novel approach to efficiently feeding knowledge to language models (LLMs) during prediction by integrating retrieval and generation processes within a unified framework. While the Retrieval-Augmented Generation (RAG) model addresses gaps in LLMs' training data and knowledge limits, it is hindered by token limit restrictions and dependency on the retrieval system's accuracy. Our proposed architecture incorporates in-context vectors (ICV) to overcome these challenges. ICV recasts in-context learning by using latent embeddings of LLMs to create a vector that captures essential task information. This vector is then used to shift the latent states of the LLM, enhancing the generation process without adding demonstration examples to the prompt. ICV directly integrates information into the model, enabling it to process this information more effectively. Our extensive experimental evaluation demonstrates that ICV outperforms standard in-context learning and fine-tuning across question-answering, information retrieval, and other tasks. This approach mitigates the limitations of current RAG models and offers a more robust solution for handling extensive and diverse datasets. Despite leveraging a fraction of the parameters, our ICV-enhanced model achieves competitive performance against models like LLaMA-3, Gemma, and Phi-3, significantly reducing computational costs and memory requirements. ICV reduces prompt length, is easy to control, surpasses token limitations, and is computationally efficient compared to fine-tuning.