InternSVG: Towards Unified SVG Tasks with Multimodal Large Language Models

2025-10-13

AI Summary

General-purpose SVG modeling has long been hindered by fragmented datasets, poor cross-task transferability, and challenges in capturing structural complexity. Method: We propose InternSVG, a unified multimodal large language model specifically designed for SVGs, introducing SVG-specific special tokens, subword-level embedding initialization, and a two-stage progressive training strategy to jointly model understanding, editing, and generation tasks. We further construct SAgoge—the first comprehensive SVG dataset covering both static graphics and dynamic animations—and SArena, its corresponding benchmark. Contribution/Results: Experiments demonstrate that InternSVG consistently outperforms existing open-source and proprietary methods across SArena and multiple mainstream benchmarks, achieving significant improvements in generalization and cross-task transfer capability. This work establishes a new paradigm for unified representation and automated processing of SVG content.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Large Vision Models
• Natural Language Processing: (Large) Language Models

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

General SVG modeling remains challenging due to fragmented datasets, limited transferability of methods across tasks, and the difficulty of handling structural complexity. In response, we leverage the strong transfer and generalization capabilities of multimodal large language models (MLLMs) to achieve unified modeling for SVG understanding, editing, and generation. We present the InternSVG family, an integrated data-benchmark-model suite. At its core is SAgoge, the largest and most comprehensive multimodal dataset for SVG tasks, encompassing both static graphics and dynamic animations. It covers icons, long-sequence illustrations, scientific diagrams, and dynamic animations, supporting tasks of varied difficulty levels and providing deeper hierarchies with richer attributes compared to previous datasets. Based on this resource, we introduce SArena, a companion benchmark with comprehensive task definitions and standardized evaluation that aligns with the domains and difficulty spectrum covered by SAgoge. Building on these foundations, we propose InternSVG, a unified MLLM for SVG understanding, editing, and generation with SVG-specific special tokens, subword-based embedding initialization, and a two-stage training strategy that progresses from short static SVGs to long-sequence illustrations and complex animations. This unified formulation induces positive transfer and improves overall performance. Experiments on SArena and prior benchmark confirm that InternSVG achieves substantial gains and consistently outperforms leading open and proprietary counterparts.

Problem

• Unifying SVG understanding, editing, and generation tasks
• Addressing fragmented datasets and limited method transferability
• Handling structural complexity in static and dynamic SVGs

Innovation

• Leveraging MLLMs for unified SVG understanding, editing, generation
• Introducing SAgoge, largest multimodal SVG dataset with animations
• Proposing InternSVG model with SVG tokens and two-stage training