Xiaohui Zeng

Yan Wang, Wenjie Luo, Junjie Bai et al. · nvidia

End-to-end architectures trained via imitation learning have advanced autonomous driving by scaling model size and data, yet performance remains brittle in safety-critical long-tail scenarios where supervision is sparse and causal understanding is limited. To address this, we introduce Alpamayo-R1 (AR1), a vision-language-action model (VLA) that integrates Chain of Causation reasoning with trajectory planning to enhance decision-making in complex driving scenarios. Our approach features three key innovations: (1) the Chain of Causation (CoC) dataset, built through a hybrid auto-labeling and human-in-the-loop pipeline producing decision-grounded, causally linked reasoning traces aligned with driving behaviors; (2) a modular VLA architecture combining Cosmos-Reason, a Vision-Language Model pre-trained for Physical AI applications, with a diffusion-based trajectory decoder that generates dynamically feasible plans in real time; (3) a multi-stage training strategy using supervised fine-tuning to elicit reasoning and reinforcement learning (RL) to optimize reasoning quality via large reasoning model feedback and enforce reasoning-action consistency. Evaluation shows AR1 achieves up to a 12% improvement in planning accuracy on challenging cases compared to a trajectory-only baseline, with a 35% reduction in off-road rate and 25% reduction in close encounter rate in closed-loop simulation. RL post-training improves reasoning quality by 45% as measured by a large reasoning model critic and reasoning-action consistency by 37%. Model scaling from 0.5B to 7B parameters shows consistent improvements. On-vehicle road tests confirm real-time performance (99 ms latency) and successful urban deployment. By bridging interpretable reasoning with precise control, AR1 demonstrates a practical path towards Level 4 autonomous driving. We plan to release AR1 models and a subset of the CoC in a future update.

Arslan Ali, Junjie Bai, Maciej Bala et al. · nvidia

We introduce [Cosmos-Predict2.5], the latest generation of the Cosmos World Foundation Models for Physical AI. Built on a flow-based architecture, [Cosmos-Predict2.5] unifies Text2World, Image2World, and Video2World generation in a single model and leverages [Cosmos-Reason1], a Physical AI vision-language model, to provide richer text grounding and finer control of world simulation. Trained on 200M curated video clips and refined with reinforcement learning-based post-training, [Cosmos-Predict2.5] achieves substantial improvements over [Cosmos-Predict1] in video quality and instruction alignment, with models released at 2B and 14B scales. These capabilities enable more reliable synthetic data generation, policy evaluation, and closed-loop simulation for robotics and autonomous systems. We further extend the family with [Cosmos-Transfer2.5], a control-net style framework for Sim2Real and Real2Real world translation. Despite being 3.5$\times$ smaller than [Cosmos-Transfer1], it delivers higher fidelity and robust long-horizon video generation. Together, these advances establish [Cosmos-Predict2.5] and [Cosmos-Transfer2.5] as versatile tools for scaling embodied intelligence. To accelerate research and deployment in Physical AI, we release source code, pretrained checkpoints, and curated benchmarks under the NVIDIA Open Model License at https://github.com/nvidia-cosmos/cosmos-predict2.5 and https://github.com/nvidia-cosmos/cosmos-transfer2.5. We hope these open resources lower the barrier to adoption and foster innovation in building the next generation of embodied intelligence.

Xuanchi Ren, Jiahui Huang, Xiaohui Zeng et al. · utoronto

We present XCube (abbreviated as $\mathcal{X}^3$), a novel generative model for high-resolution sparse 3D voxel grids with arbitrary attributes. Our model can generate millions of voxels with a finest effective resolution of up to $1024^3$ in a feed-forward fashion without time-consuming test-time optimization. To achieve this, we employ a hierarchical voxel latent diffusion model which generates progressively higher resolution grids in a coarse-to-fine manner using a custom framework built on the highly efficient VDB data structure. Apart from generating high-resolution objects, we demonstrate the effectiveness of XCube on large outdoor scenes at scales of 100m$\times$100m with a voxel size as small as 10cm. We observe clear qualitative and quantitative improvements over past approaches. In addition to unconditional generation, we show that our model can be used to solve a variety of tasks such as user-guided editing, scene completion from a single scan, and text-to-3D. The source code and more results can be found at https://research.nvidia.com/labs/toronto-ai/xcube/.

Xiaohui Zeng, Raquel Urtasun, Richard Zemel et al.

In this paper, we present a non-parametric structured latent variable model for image generation, called NP-DRAW, which sequentially draws on a latent canvas in a part-by-part fashion and then decodes the image from the canvas. Our key contributions are as follows. 1) We propose a non-parametric prior distribution over the appearance of image parts so that the latent variable ``what-to-draw'' per step becomes a categorical random variable. This improves the expressiveness and greatly eases the learning compared to Gaussians used in the literature. 2) We model the sequential dependency structure of parts via a Transformer, which is more powerful and easier to train compared to RNNs used in the literature. 3) We propose an effective heuristic parsing algorithm to pre-train the prior. Experiments on MNIST, Omniglot, CIFAR-10, and CelebA show that our method significantly outperforms previous structured image models like DRAW and AIR and is competitive to other generic generative models. Moreover, we show that our model's inherent compositionality and interpretability bring significant benefits in the low-data learning regime and latent space editing. Code is available at https://github.com/ZENGXH/NPDRAW.

Xiaohui Zeng, Renjie Liao, Li Gu et al.

In this paper, we propose the differentiable mask-matching network (DMM-Net) for solving the video object segmentation problem where the initial object masks are provided. Relying on the Mask R-CNN backbone, we extract mask proposals per frame and formulate the matching between object templates and proposals at one time step as a linear assignment problem where the cost matrix is predicted by a CNN. We propose a differentiable matching layer by unrolling a projected gradient descent algorithm in which the projection exploits the Dykstra's algorithm. We prove that under mild conditions, the matching is guaranteed to converge to the optimum. In practice, it performs similarly to the Hungarian algorithm during inference. Meanwhile, we can back-propagate through it to learn the cost matrix. After matching, a refinement head is leveraged to improve the quality of the matched mask. Our DMM-Net achieves competitive results on the largest video object segmentation dataset YouTube-VOS. On DAVIS 2017, DMM-Net achieves the best performance without online learning on the first frames. Without any fine-tuning, DMM-Net performs comparably to state-of-the-art methods on SegTrack v2 dataset. At last, our matching layer is very simple to implement; we attach the PyTorch code ($<50$ lines) in the supplementary material. Our code is released at https://github.com/ZENGXH/DMM_Net.

Kevin Xie, Jonathan Lorraine, Tianshi Cao et al. · nvidia, utoronto

Recent text-to-3D generation approaches produce impressive 3D results but require time-consuming optimization that can take up to an hour per prompt. Amortized methods like ATT3D optimize multiple prompts simultaneously to improve efficiency, enabling fast text-to-3D synthesis. However, they cannot capture high-frequency geometry and texture details and struggle to scale to large prompt sets, so they generalize poorly. We introduce LATTE3D, addressing these limitations to achieve fast, high-quality generation on a significantly larger prompt set. Key to our method is 1) building a scalable architecture and 2) leveraging 3D data during optimization through 3D-aware diffusion priors, shape regularization, and model initialization to achieve robustness to diverse and complex training prompts. LATTE3D amortizes both neural field and textured surface generation to produce highly detailed textured meshes in a single forward pass. LATTE3D generates 3D objects in 400ms, and can be further enhanced with fast test-time optimization.

Jiawei Ren, Kevin Xie, Ashkan Mirzaei et al.

We present L4GM, the first 4D Large Reconstruction Model that produces animated objects from a single-view video input -- in a single feed-forward pass that takes only a second. Key to our success is a novel dataset of multiview videos containing curated, rendered animated objects from Objaverse. This dataset depicts 44K diverse objects with 110K animations rendered in 48 viewpoints, resulting in 12M videos with a total of 300M frames. We keep our L4GM simple for scalability and build directly on top of LGM, a pretrained 3D Large Reconstruction Model that outputs 3D Gaussian ellipsoids from multiview image input. L4GM outputs a per-frame 3D Gaussian Splatting representation from video frames sampled at a low fps and then upsamples the representation to a higher fps to achieve temporal smoothness. We add temporal self-attention layers to the base LGM to help it learn consistency across time, and utilize a per-timestep multiview rendering loss to train the model. The representation is upsampled to a higher framerate by training an interpolation model which produces intermediate 3D Gaussian representations. We showcase that L4GM that is only trained on synthetic data generalizes extremely well on in-the-wild videos, producing high quality animated 3D assets.

Bowen Chen, Huan Ling, Xiaohui Zeng et al.

Manually labeling video datasets for segmentation tasks is extremely time consuming. In this paper, we introduce ScribbleBox, a novel interactive framework for annotating object instances with masks in videos. In particular, we split annotation into two steps: annotating objects with tracked boxes, and labeling masks inside these tracks. We introduce automation and interaction in both steps. Box tracks are annotated efficiently by approximating the trajectory using a parametric curve with a small number of control points which the annotator can interactively correct. Our approach tolerates a modest amount of noise in the box placements, thus typically only a few clicks are needed to annotate tracked boxes to a sufficient accuracy. Segmentation masks are corrected via scribbles which are efficiently propagated through time. We show significant performance gains in annotation efficiency over past work. We show that our ScribbleBox approach reaches 88.92% J&F on DAVIS2017 with 9.14 clicks per box track, and 4 frames of scribble annotation.

Yu-Siang Wang, Chenxi Liu, Xiaohui Zeng et al.

In this paper, we study the problem of parsing structured knowledge graphs from textual descriptions. In particular, we consider the scene graph representation that considers objects together with their attributes and relations: this representation has been proved useful across a variety of vision and language applications. We begin by introducing an alternative but equivalent edge-centric view of scene graphs that connect to dependency parses. Together with a careful redesign of label and action space, we combine the two-stage pipeline used in prior work (generic dependency parsing followed by simple post-processing) into one, enabling end-to-end training. The scene graphs generated by our learned neural dependency parser achieve an F-score similarity of 49.67% to ground truth graphs on our evaluation set, surpassing best previous approaches by 5%. We further demonstrate the effectiveness of our learned parser on image retrieval applications.

Xiaohui Zeng, Chenxi Liu, Yu-Siang Wang et al.

Generating adversarial examples is an intriguing problem and an important way of understanding the working mechanism of deep neural networks. Most existing approaches generated perturbations in the image space, i.e., each pixel can be modified independently. However, in this paper we pay special attention to the subset of adversarial examples that correspond to meaningful changes in 3D physical properties (like rotation and translation, illumination condition, etc.). These adversaries arguably pose a more serious concern, as they demonstrate the possibility of causing neural network failure by easy perturbations of real-world 3D objects and scenes. In the contexts of object classification and visual question answering, we augment state-of-the-art deep neural networks that receive 2D input images with a rendering module (either differentiable or not) in front, so that a 3D scene (in the physical space) is rendered into a 2D image (in the image space), and then mapped to a prediction (in the output space). The adversarial perturbations can now go beyond the image space, and have clear meanings in the 3D physical world. Though image-space adversaries can be interpreted as per-pixel albedo change, we verify that they cannot be well explained along these physically meaningful dimensions, which often have a non-local effect. But it is still possible to successfully attack beyond the image space on the physical space, though this is more difficult than image-space attacks, reflected in lower success rates and heavier perturbations required.