SEED-Story：基于大语言模型的多模态长故事生成
SEED-Story: Multimodal Long Story Generation with Large Language Model
AI总结：
提出SEED-Story方法，利用MLLM生成长达25序列的多模态故事，并引入多模态注意力汇聚机制和StoryStream数据集。
摘要：
随着图像生成和开放式文本生成的显著进步，交错图文内容的创作已成为一个日益引人入胜的领域。多模态故事生成以交错方式生成叙事文本和生动图像为特征，已发展为一项具有广泛应用的极具价值且实用的任务。然而，该任务面临重大挑战，因为它需要理解文本与图像之间复杂的相互作用，并能够生成长序列的连贯且上下文相关的文本和视觉内容。在这项工作中，我们提出了 SEED-Story，一种利用多模态大语言模型（MLLM）来生成扩展多模态故事的新方法。我们的模型建立在 MLLM 强大的理解能力之上，预测文本标记以及视觉标记，随后使用适配的视觉去标记器进行处理，以生成具有一致角色和风格的图像。我们进一步提出了多模态注意力汇聚机制，使得能够以高度高效的自回归方式生成多达 25 个序列（训练时仅为 10 个）的故事。此外，我们提出了一个名为 StoryStream 的大规模高分辨率数据集，用于训练我们的模型，并从多个方面对多模态故事生成任务进行定量评估。
英文摘要：
With the remarkable advancements in image generation and open-form text generation, the creation of interleaved image-text content has become an increasingly intriguing field. Multimodal story generation, characterized by producing narrative texts and vivid images in an interleaved manner, has emerged as a valuable and practical task with broad applications. However, this task poses significant challenges, as it necessitates the comprehension of the complex interplay between texts and images, and the ability to generate long sequences of coherent, contextually relevant texts and visuals. In this work, we propose SEED-Story, a novel method that leverages a Multimodal Large Language Model (MLLM) to generate extended multimodal stories. Our model, built upon the powerful comprehension capability of MLLM, predicts text tokens as well as visual tokens, which are subsequently processed with an adapted visual de-tokenizer to produce images with consistent characters and styles. We further propose multimodal attention sink mechanism to enable the generation of stories with up to 25 sequences (only 10 for training) in a highly efficient autoregressive manner. Additionally, we present a large-scale and high-resolution dataset named StoryStream for training our model and quantitatively evaluating the task of multimodal story generation in various aspects.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
