Enregistré dans:
Détails bibliographiques
Auteurs principaux: Raaijmakers, Stephan, Bakker, Roos, Cremers, Anita, de Kleijn, Roy, Kouwenhoven, Tom, Verhoef, Tessa
Format: Preprint
Publié: 2024
Sujets:
Accès en ligne:https://arxiv.org/abs/2402.19218
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866929259486904320
author Raaijmakers, Stephan
Bakker, Roos
Cremers, Anita
de Kleijn, Roy
Kouwenhoven, Tom
Verhoef, Tessa
author_facet Raaijmakers, Stephan
Bakker, Roos
Cremers, Anita
de Kleijn, Roy
Kouwenhoven, Tom
Verhoef, Tessa
contents 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 for addressing this memory. We add this augmented memory to a Generative Adversarial Network-inspired Transformer architecture. This setup allows for implementing arbitrary felicity conditions on the generated language of the Transformer. We first demonstrate how this machinery can be deployed for handling factual questions in goal-oriented dialogues. Secondly, we demonstrate that our approach can be useful for applications like {\it style adaptation} as well: the adaptation of utterances according to certain stylistic (external) constraints, like social properties of human interlocutors in dialogues.
format Preprint
id arxiv_https___arxiv_org_abs_2402_19218
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Memory-Augmented Generative Adversarial Transformers
Raaijmakers, Stephan
Bakker, Roos
Cremers, Anita
de Kleijn, Roy
Kouwenhoven, Tom
Verhoef, Tessa
Computation and Language
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 for addressing this memory. We add this augmented memory to a Generative Adversarial Network-inspired Transformer architecture. This setup allows for implementing arbitrary felicity conditions on the generated language of the Transformer. We first demonstrate how this machinery can be deployed for handling factual questions in goal-oriented dialogues. Secondly, we demonstrate that our approach can be useful for applications like {\it style adaptation} as well: the adaptation of utterances according to certain stylistic (external) constraints, like social properties of human interlocutors in dialogues.
title Memory-Augmented Generative Adversarial Transformers
topic Computation and Language
url https://arxiv.org/abs/2402.19218