标题： LLM-I：大语言模型是自然交错的多模态创作者 显示英文标题

标题： LLM-I: LLMs are Naturally Interleaved Multimodal Creators

摘要： 我们提出LLM-Interleaved（LLM-I），一个灵活且动态的框架，将交错的图像文本生成重新表述为工具使用问题。 LLM-I旨在克服当前统一模型的“单一工具”瓶颈，这些模型仅限于合成图像，并在需要事实依据或程序精度的任务上表现不佳。 我们的框架使一个中心LLM或MLLM代理能够智能地协调多种专业视觉工具，包括在线图像搜索、基于扩散的生成、代码执行和图像编辑。 该代理通过一种结合基于规则的逻辑与LLM和MLLM评估者的判断的混合奖励系统的强化学习（RL）框架进行训练，以熟练选择和应用这些工具。 在使用四种不同模型主干的新数据集上进行训练，LLM-I表现出最先进的性能，在四个基准测试中大幅优于现有方法。 我们还引入了一种新的测试时缩放策略，进一步提升了性能。 项目页面：https://github.com/ByteDance-BandAI/LLM-I. 显示英文摘要

摘要： We propose LLM-Interleaved (LLM-I), a flexible and dynamic framework that reframes interleaved image-text generation as a tool-use problem. LLM-I is designed to overcome the "one-tool" bottleneck of current unified models, which are limited to synthetic imagery and struggle with tasks requiring factual grounding or programmatic precision. Our framework empowers a central LLM or MLLM agent to intelligently orchestrate a diverse toolkit of specialized visual tools, including online image search, diffusion-based generation, code execution, and image editing. The agent is trained to select and apply these tools proficiently via a Reinforcement Learning (RL) framework that features a hybrid reward system combining rule-based logic with judgments from LLM and MLLM evaluators. Trained on a diverse new dataset using four different model backbones, LLM-I demonstrates state-of-the-art performance, outperforming existing methods by a large margin across four benchmarks. We also introduce a novel test-time scaling strategy that provides further performance gains. Project Page: https://github.com/ByteDance-BandAI/LLM-I.