Memory-Augmented Generative Adversarial Transformers

arXiv:2402.19218v1 Announce Type: new
Abstract: Conversational AI systems that rely on Large Language Models, like Transformers, have difficulty interweaving external data (like facts) with the language they generate. Vanilla Transformer architectures are not designed for answering factual questions with high accuracy. This paper investigates a possible route for addressing this problem. We propose to extend the standard Transformer architecture with an additional memory bank holding extra information (such as facts drawn from a knowledge base), and an extra attention layer …

abstract accuracy adversarial ai systems architectures arxiv conversational conversational ai cs.cl data external data facts generate generative language language models large language large language models memory paper questions route systems transformer transformers type vanilla transformer