MAGID: An Automated Pipeline for Generating Synthetic Multi-modal Datasets

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Aboutalebi, Hossein, Song, Hwanjun, Xie, Yusheng, Gupta, Arshit, Sun, Justin, Su, Hang, Shalyminov, Igor, Pappas, Nikolaos, Singh, Siffi, Mansour, Saab
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929525198159872
author Aboutalebi, Hossein
Song, Hwanjun
Xie, Yusheng
Gupta, Arshit
Sun, Justin
Su, Hang
Shalyminov, Igor
Pappas, Nikolaos
Singh, Siffi
Mansour, Saab
author_facet Aboutalebi, Hossein
Song, Hwanjun
Xie, Yusheng
Gupta, Arshit
Sun, Justin
Su, Hang
Shalyminov, Igor
Pappas, Nikolaos
Singh, Siffi
Mansour, Saab
contents Development of multimodal interactive systems is hindered by the lack of rich, multimodal (text, images) conversational data, which is needed in large quantities for LLMs. Previous approaches augment textual dialogues with retrieved images, posing privacy, diversity, and quality constraints. In this work, we introduce Multimodal Augmented Generative Images Dialogues (MAGID), a framework to augment text-only dialogues with diverse and high-quality images. Subsequently, a diffusion model is applied to craft corresponding images, ensuring alignment with the identified text. Finally, MAGID incorporates an innovative feedback loop between an image description generation module (textual LLM) and image quality modules (addressing aesthetics, image-text matching, and safety), that work in tandem to generate high-quality and multi-modal dialogues. We compare MAGID to other SOTA baselines on three dialogue datasets, using automated and human evaluation. Our results show that MAGID is comparable to or better than baselines, with significant improvements in human evaluation, especially against retrieval baselines where the image database is small.
format Preprint
id arxiv_https___arxiv_org_abs_2403_03194
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle MAGID: An Automated Pipeline for Generating Synthetic Multi-modal Datasets
Aboutalebi, Hossein
Song, Hwanjun
Xie, Yusheng
Gupta, Arshit
Sun, Justin
Su, Hang
Shalyminov, Igor
Pappas, Nikolaos
Singh, Siffi
Mansour, Saab
Computation and Language
Development of multimodal interactive systems is hindered by the lack of rich, multimodal (text, images) conversational data, which is needed in large quantities for LLMs. Previous approaches augment textual dialogues with retrieved images, posing privacy, diversity, and quality constraints. In this work, we introduce Multimodal Augmented Generative Images Dialogues (MAGID), a framework to augment text-only dialogues with diverse and high-quality images. Subsequently, a diffusion model is applied to craft corresponding images, ensuring alignment with the identified text. Finally, MAGID incorporates an innovative feedback loop between an image description generation module (textual LLM) and image quality modules (addressing aesthetics, image-text matching, and safety), that work in tandem to generate high-quality and multi-modal dialogues. We compare MAGID to other SOTA baselines on three dialogue datasets, using automated and human evaluation. Our results show that MAGID is comparable to or better than baselines, with significant improvements in human evaluation, especially against retrieval baselines where the image database is small.
title MAGID: An Automated Pipeline for Generating Synthetic Multi-modal Datasets
topic Computation and Language
url https://arxiv.org/abs/2403.03194