MAGI-1实现大规模自回归视频生成，支持长时一致性与实时流式输出。
MAGI-1: Autoregressive Video Generation at Scale
- 通过逐块自回归预测视频片段，实现因果时间建模。
- 最大模型达240亿参数，支持400万词元上下文长度。
- 支持分块提示控制，推理内存恒定，适合实时部署。
我们提出MAGI-1，一种通过自回归方式逐块预测连续帧序列的世界模型。该模型在训练中学习消除随时间单调增加的每块噪声，实现因果时序建模，并天然支持流式生成。MAGI-1在文本条件图像到视频（I2V）任务中表现优异，具备高时间一致性和可扩展性，这得益于多项算法创新和专用基础设施。通过分块提示实现可控生成，且推理峰值内存不随视频长度增加，支持实时、低内存消耗部署。最大版本MAGI-1包含240亿参数，支持最高400万词元的上下文长度，验证了方法的可扩展性与鲁棒性。代码与模型开源于https://github.com/SandAI-org/MAGI-1 和 https://github.com/SandAI-org/MagiAttention，产品可访问 https://sand.ai。
原文摘要 · Abstract (English)
We present MAGI-1, a world model that generates videos by autoregressively predicting a sequence of video chunks, defined as fixed-length segments of consecutive frames. Trained to denoise per-chunk noise that increases monotonically over time, MAGI-1 enables causal temporal modeling and naturally supports streaming generation. It achieves strong performance on image-to-video (I2V) tasks conditioned on text instructions, providing high temporal consistency and scalability, which are made possible by several algorithmic innovations and a dedicated infrastructure stack. MAGI-1 facilitates controllable generation via chunk-wise prompting and supports real-time, memory-efficient deployment by maintaining constant peak inference cost, regardless of video length. The largest variant of MAGI-1 comprises 24 billion parameters and supports context lengths of up to 4 million tokens, demonstrating the scalability and robustness of our approach. The code and models are available at https://github.com/SandAI-org/MAGI-1 and https://github.com/SandAI-org/MagiAttention. The product can be accessed at https://sand.ai.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。