Video Foundation Models (VFMs) have recently been used to simulate the real world to train physical AI systems and develop creative visual experiences. However, there are significant challenges in training large-scale, high quality VFMs that can generate high-quality videos. We present a scalable, open-source VFM training pipeline with NVIDIA NeMo, providing accelerated video dataset curation, multimodal data loading, and parallelized video diffusion model training and inference. We also provide a comprehensive performance analysis highlighting best practices for efficient VFM training and inference.

Physical AI needs to be trained digitally first. It needs a digital twin of itself, the policy model, and a digital twin of the world, the world model. In this paper, we present the Cosmos World Foundation Model Platform to help developers build customized world models for their Physical AI setups. We position a world foundation model as a general-purpose world model that can be fine-tuned into customized world models for downstream applications. Our platform covers a video curation pipeline, pre-trained world foundation models, examples of post-training of pre-trained world foundation models, and video tokenizers. To help Physical AI builders solve the most critical problems of our society, we make our platform open-source and our models open-weight with permissive licenses available via https://github.com/NVIDIA/Cosmos.

Recent techniques in deep generative modeling have leveraged Markov generative processes to learn complex, high-dimensional probability distributions in a more structured and flexible manner [17]. By integrating Markov chain methods with deep neural architectures, these approaches aim to exploit the representational power of deep networks while maintaining a tractable and theoretically grounded training procedure. In contrast to early generative models that relied heavily on direct maximum likelihood estimation or adversarial objectives, this class of methods employs iterative stochastic transformations--often expressed as Markovian updates--to gradually refine initial noise samples into samples drawn from the desired target distribution. Diffusion and flow matching models represent two prominent classes of generative approaches that construct data samples through a sequence of continuous transformations. Diffusion models [6, 13] introduce a forward-noising and reverse-denoising process, progressively refining a simple noise distribution into a complex target distribution by learning to undo incremental noise corruption at each step.

In this paper, we argue that iterative computation with diffusion models offers a powerful paradigm for not only generation but also visual perception tasks. We unify tasks such as depth estimation, optical flow, and amodal segmentation under the framework of image-to-image translation, and show how diffusion models benefit from scaling training and test-time compute for these perceptual tasks. Through a careful analysis of these scaling properties, we formulate computeoptimal training and inference recipes to scale diffusion models for visual perception tasks. Our models achieve competitive performance to state-of-the-art methods using significantly less data and compute. Diffusion models have emerged as powerful techniques for generating images and videos, while showing excellent scaling behaviors. In this paper, we present a unified framework to perform a variety of perceptual tasks -- depth estimation, optical flow estimation, and amodal segmentation -- with a single diffusion model, as illustrated in Figure 1. Previous works such as Marigold (Ke et al., 2024), FlowDiffuser (Luo et al., 2024), and pix2gestalt (Ozguroglu et al., 2024) demonstrate the potential of repurposing image diffusion models for various inverse vision tasks individually.

Automated long-form story generation typically employs long-context large language models (LLMs) for one-shot creation, which can produce cohesive but not necessarily engaging content. We introduce Storytelling With Action Guidance (SWAG), a novel approach to storytelling with LLMs. Our approach reduces story writing to a search problem through a two-model feedback loop: one LLM generates story content, and another auxiliary LLM is used to choose the next best "action" to steer the story's future direction. Our results show that SWAG can substantially outperform previous end-to-end story generation techniques when evaluated by GPT-4 and through human evaluation, and our SWAG pipeline using only open-source models surpasses GPT-3.5-Turbo.