A formal theory proves model exploitation is essentially unavoidable on large policy sets in RL, generalizes reward hacking results, and derives a safe horizon for a relaxed version of exploitation.
super hub Canonical reference
Cosmos World Foundation Model Platform for Physical AI
Canonical reference. 79% of citing Pith papers cite this work as background.
145
Pith papers citing it
Background
79% of classified citations
abstract
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 Cosmos open-source and our models open-weight with permissive licenses available via https://github.com/nvidia-cosmos/cosmos-predict1.
hub tools
citation-role summary
background 39
method 6
baseline 3
citation-polarity summary
claims ledger
- abstract 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,
authors
co-cited works
representative citing papers
AnyFlow enables any-step video diffusion by distilling flow-map transitions over arbitrary time intervals with on-policy backward simulation.
SimWorld Studio deploys an evolving coding agent to create adaptive 3D environments that co-evolve with embodied learners, delivering 18-point success-rate gains over fixed environments in navigation benchmarks.
Physics-IQ benchmark reveals that generative video models exhibit limited physical understanding unrelated to their visual quality.
World models introduce a stealthy poisoning vector into robot learning pipelines where malicious prompts or dynamics in teleoperated data activate only during synthetic trajectory generation, enabling backdoors in downstream policies.
JOPAT jointly models pixels, point tracks, and actions in a diffusion transformer and reports gains over pixel-only baselines on long-horizon robot tasks with occlusion and off-screen motion.
Aero-World adapts a pretrained latent diffusion transformer for action-conditioned aerial video generation by injecting inertial action tokens and using a frozen latent-space Physics Probe for inertial consistency supervision during LoRA finetuning, with a new AeroBench benchmark showing improved AA
BandTok tokenizes Mel-spectrograms as independent time-frequency band tokens from a single codebook and pairs it with 2D RoPE in an autoregressive model to improve music generation over residual multi-codebook tokenizers.
CreFlow combines LTL compositional rewards with credit-aware NFT and corrective reflow losses in online RL to improve embodied video diffusion models, raising downstream task success by 23.8 percentage points on eight bimanual manipulation tasks.
A GenAI-based method extracts representations from unstructured data and uses a neural network to fit marginal structural models that recover causal effects of treatment feature sequences including their positions.
NoiseGate learns per-latent timestep schedules as an information-gating policy in diffusion-based world action models, yielding consistent gains on RoboTwin manipulation tasks.
LiVeAction is a lightweight asymmetric neural codec using an FFT-inspired encoder and variance-based training that outperforms generative tokenizers in rate-distortion while supporting real-time use on resource-constrained sensors across modalities.
EA-WM generates more accurate robot world rollouts by projecting actions as structured visual fields in camera space and using event-aware bidirectional fusion to better capture interaction dynamics.
Action Agent pairs LLM-driven video generation with a flow-constrained diffusion transformer to produce velocity commands, raising video success to 86% and delivering 64.7% real-world navigation on a Unitree G1 humanoid.
Being-H0.7 adds future-aware latent reasoning to direct VLA policies via dual-branch alignment on latent queries, matching world-model benefits at VLA efficiency.
OmniShotCut treats shot boundary detection as structured relational prediction via a shot-query Transformer, uses fully synthetic transitions for training data, and releases OmniShotCutBench for evaluation.
VistaBot integrates 4D geometry estimation and spatiotemporal view synthesis into action policies to improve cross-view generalization by 2.6-2.8x on a new VGS metric in simulation and real tasks.
RoboWM-Bench evaluates video world models by converting their manipulation video predictions into executable actions validated in simulation, showing that visual plausibility does not guarantee physical executability.
EgoTL provides a new egocentric dataset with think-aloud chains and metric labels that benchmarks VLMs on long-horizon tasks and improves their planning, reasoning, and spatial grounding after finetuning.
ViVa turns a video generator into a value model for robot RL that jointly forecasts future states and task value, yielding better performance on real-world box assembly when integrated with RECAP.
MoRight disentangles object and camera motion via canonical-view specification and temporal cross-view attention, while decomposing motion into active user-driven and passive consequence components to learn and apply causality in video generation.
SEM-ROVER generates large multiview-consistent 3D urban driving scenes via semantic-conditioned diffusion on Σ-Voxfield voxel grids with progressive outpainting and deferred rendering.
Delta tokens compress VFM feature differences into single tokens, enabling a lightweight generative world model that predicts diverse futures with far lower compute than existing approaches.
KITScenes LongTail supplies multimodal driving data and multilingual expert reasoning traces to benchmark models on rare scenarios beyond basic safety metrics.
citing papers explorer
-
Imperfect World Models are Exploitable
A formal theory proves model exploitation is essentially unavoidable on large policy sets in RL, generalizes reward hacking results, and derives a safe horizon for a relaxed version of exploitation.
-
AnyFlow: Any-Step Video Diffusion Model with On-Policy Flow Map Distillation
AnyFlow enables any-step video diffusion by distilling flow-map transitions over arbitrary time intervals with on-policy backward simulation.
-
SimWorld Studio: Automatic Environment Generation with Evolving Coding Agent for Embodied Agent Learning
SimWorld Studio deploys an evolving coding agent to create adaptive 3D environments that co-evolve with embodied learners, delivering 18-point success-rate gains over fixed environments in navigation benchmarks.
-
Do generative video models understand physical principles?
Physics-IQ benchmark reveals that generative video models exhibit limited physical understanding unrelated to their visual quality.
-
Targeting World Models to Compromise Robot Learning Pipelines
World models introduce a stealthy poisoning vector into robot learning pipelines where malicious prompts or dynamics in teleoperated data activate only during synthetic trajectory generation, enabling backdoors in downstream policies.
-
Point Tracking Improves World Action Models
JOPAT jointly models pixels, point tracks, and actions in a diffusion transformer and reports gains over pixel-only baselines on long-horizon robot tasks with occlusion and off-screen motion.
-
Aero-World: Action-Conditioned Aerial Video Generation from Inertial Controls
Aero-World adapts a pretrained latent diffusion transformer for action-conditioned aerial video generation by injecting inertial action tokens and using a frozen latent-space Physics Probe for inertial consistency supervision during LoRA finetuning, with a new AeroBench benchmark showing improved AA
-
Modeling Music as a Time-Frequency Image: A 2D Tokenizer for Music Generation
BandTok tokenizes Mel-spectrograms as independent time-frequency band tokens from a single codebook and pairs it with 2D RoPE in an autoregressive model to improve music generation over residual multi-codebook tokenizers.
-
CreFlow: Corrective Reflow for Sparse-Reward Embodied Video Diffusion RL
CreFlow combines LTL compositional rewards with credit-aware NFT and corrective reflow losses in online RL to improve embodied video diffusion models, raising downstream task success by 23.8 percentage points on eight bimanual manipulation tasks.
-
GenAI Powered Dynamic Causal Inference with Unstructured Data
A GenAI-based method extracts representations from unstructured data and uses a neural network to fit marginal structural models that recover causal effects of treatment feature sequences including their positions.
-
NoiseGate: Learning Per-Latent Timestep Schedules as Information Gating in World Action Models
NoiseGate learns per-latent timestep schedules as an information-gating policy in diffusion-based world action models, yielding consistent gains on RoboTwin manipulation tasks.
-
LiVeAction: a Lightweight, Versatile, and Asymmetric Neural Codec Design for Real-time Operation
LiVeAction is a lightweight asymmetric neural codec using an FFT-inspired encoder and variance-based training that outperforms generative tokenizers in rate-distortion while supporting real-time use on resource-constrained sensors across modalities.
-
EA-WM: Event-Aware Generative World Model with Structured Kinematic-to-Visual Action Fields
EA-WM generates more accurate robot world rollouts by projecting actions as structured visual fields in camera space and using event-aware bidirectional fusion to better capture interaction dynamics.
-
Action Agent: Agentic Video Generation Meets Flow-Constrained Diffusion
Action Agent pairs LLM-driven video generation with a flow-constrained diffusion transformer to produce velocity commands, raising video success to 86% and delivering 64.7% real-world navigation on a Unitree G1 humanoid.
-
Being-H0.7: A Latent World-Action Model from Egocentric Videos
Being-H0.7 adds future-aware latent reasoning to direct VLA policies via dual-branch alignment on latent queries, matching world-model benefits at VLA efficiency.
-
OmniShotCut: Holistic Relational Shot Boundary Detection with Shot-Query Transformer
OmniShotCut treats shot boundary detection as structured relational prediction via a shot-query Transformer, uses fully synthetic transitions for training data, and releases OmniShotCutBench for evaluation.
-
VistaBot: View-Robust Robot Manipulation via Spatiotemporal-Aware View Synthesis
VistaBot integrates 4D geometry estimation and spatiotemporal view synthesis into action policies to improve cross-view generalization by 2.6-2.8x on a new VGS metric in simulation and real tasks.
-
RoboWM-Bench: A Benchmark for Evaluating World Models in Robotic Manipulation
RoboWM-Bench evaluates video world models by converting their manipulation video predictions into executable actions validated in simulation, showing that visual plausibility does not guarantee physical executability.
-
EgoTL: Egocentric Think-Aloud Chains for Long-Horizon Tasks
EgoTL provides a new egocentric dataset with think-aloud chains and metric labels that benchmarks VLMs on long-horizon tasks and improves their planning, reasoning, and spatial grounding after finetuning.
-
ViVa: A Video-Generative Value Model for Robot Reinforcement Learning
ViVa turns a video generator into a value model for robot RL that jointly forecasts future states and task value, yielding better performance on real-world box assembly when integrated with RECAP.
-
MoRight: Motion Control Done Right
MoRight disentangles object and camera motion via canonical-view specification and temporal cross-view attention, while decomposing motion into active user-driven and passive consequence components to learn and apply causality in video generation.
-
SEM-ROVER: Semantic Voxel-Guided Diffusion for Large-Scale Driving Scene Generation
SEM-ROVER generates large multiview-consistent 3D urban driving scenes via semantic-conditioned diffusion on Σ-Voxfield voxel grids with progressive outpainting and deferred rendering.
-
A Frame is Worth One Token: Efficient Generative World Modeling with Delta Tokens
Delta tokens compress VFM feature differences into single tokens, enabling a lightweight generative world model that predicts diverse futures with far lower compute than existing approaches.
-
LongTail Driving Scenarios with Reasoning Traces: The KITScenes LongTail Dataset
KITScenes LongTail supplies multimodal driving data and multilingual expert reasoning traces to benchmark models on rare scenarios beyond basic safety metrics.
-
PlayWorld: Learning Robot World Models from Autonomous Play
PlayWorld learns high-fidelity robot world models from unsupervised self-play, producing physically consistent video predictions that outperform models trained on human data and enabling 65% better real-world policy performance via model-based RL.
-
SPIRAL: Self-Evolving Action-Conditioned Video Generation via Reflective Planning Agents
SPIRAL is a closed-loop think-act-reflect framework using PlanAgent, VideoGenerator, and CriticAgent plus GRPO self-evolution to improve long-horizon action-conditioned video generation, with new dataset and benchmark showing gains over open-loop baselines.
-
PhysMem: Scaling Test-Time Memory for Embodied Physical Reasoning
PhysMem enables VLM-based robot planners to learn and verify physical properties through test-time interaction and hypothesis testing, raising success on a brick insertion task from 23% to 76%.
-
CoMoVi: Co-Generation of 3D Human Motions and Realistic Videos
CoMoVi co-generates 3D human motions and 2D videos synchronously in a single diffusion denoising loop using 3D-to-2D projection and dual-branch diffusion with 3D-2D cross attentions.
-
Large Video Planner Enables Generalizable Robot Control
A video foundation model trained on human demonstrations generates zero-shot plans that convert to executable robot actions on novel scenes and tasks.
-
MVAD: A Benchmark Dataset for Multimodal AI-Generated Video-Audio Detection
MVAD is the first comprehensive benchmark dataset for AI-generated multimodal video-audio detection, with three realistic forgery patterns, high-quality outputs from state-of-the-art models, and diversity across visual styles and content categories.
-
Target-Bench: Can Video World Models Achieve Mapless Path Planning with Semantic Targets?
Target-Bench shows the best off-the-shelf video world model scores only 0.341 on semantic target-approaching and directional consistency, with fine-tuning on a small robot dataset yielding measurable gains.
-
EgoWalk: A Multimodal Dataset for Robot Navigation in the Wild
EgoWalk supplies 50 hours of real-world multimodal human navigation data in varied indoor/outdoor settings together with open pipelines that auto-generate language goal annotations and traversability masks.
-
SpatialScore: Towards Comprehensive Evaluation for Spatial Intelligence
Presents SpatialScore benchmark for MLLM spatial reasoning, evaluates 49 models showing large human gap, and supplies SpatialCorpus plus SpatialAgent to improve performance.
-
DreamGen: Unlocking Generalization in Robot Learning through Video World Models
DreamGen trains robot policies on synthetic trajectories from adapted video world models, enabling a humanoid robot to perform 22 new behaviors in seen and unseen environments from a single pick-and-place teleoperation dataset.
-
EgoDex: Learning Dexterous Manipulation from Large-Scale Egocentric Video
EgoDex delivers the largest egocentric dataset with native 3D hand tracking for dexterous manipulation, enabling imitation learning policies for hand trajectory prediction on 194 tasks.
-
Lighting-Consistent Object Transfer Across Radiance Fields
Diffusion-based per-view harmonization for lighting-consistent object transfer between 3DGS scenes, using heterogeneous training data and final 3D consolidation.
-
SVI-Bench: A Dynamic Microworld for Strategic Video Intelligence
SVI-Bench is a 35K-hour sports video benchmark with 9 tasks across four cognitive pillars that reveals multimodal models drop from ~73% on action QA to 5% on agentic evidence-gathering tasks.
-
LaMo: Self-Supervised Latent Motion Priors for Physical Realism in Video Generation
LaMo adds self-supervised latent motion priors via a motion drift loss during training and motion prior guidance during sampling to boost physical fidelity in video diffusion models like CogVideoX.
-
SCOPE: Simulating Cross-game Operations in Playable Environments for FPS World Models
SCOPE adds per-pixel action conditioning to pretrained video diffusion models and releases the CrossFPS multi-game dataset to support cross-game FPS world model simulation with zero-shot transfer.
-
Accelerating Video Inverse Problem Solvers with Autoregressive Diffusion Models
AVIS applies autoregressive diffusion models to video inverse problems by streaming restoration with measurement-consistent initialization, reducing latency from 114s to 4s and raising throughput to 1.18 FPS (or 5.91 FPS in the Flash variant).
-
How You Move Tells What You'll Do: Trajectory-Conditioned Egocentric Prediction
TrajPilot predicts candidate future trajectories from egocentric context and uses them to condition action prediction in an embedding space, outperforming VLM and planner baselines on Ego-Exo4D, Ego4D, and other datasets with gains increasing at longer horizons.
-
World-Ego Modeling for Long-Horizon Evolution in Hybrid Embodied Tasks
Proposes World-Ego Modeling with WEM using CP-MoE diffusion and a new HTEWorld benchmark, claiming SOTA on hybrid navigation-manipulation tasks.
-
Latent Video Prediction Learns Better World Models
Latent prediction video models exhibit a distinct robustness profile across corruption, occlusion, fine-grained discrimination, and temporal sensitivity compared to other self-supervised video models when used as world models.
-
PanoWorld: Geometry-Consistent Panoramic Video World Modeling
PanoWorld adds depth consistency and trajectory consistency losses plus spherical adaptations to a pre-trained video model, plus a new PanoGeo dataset, to produce geometry-consistent 360 video.
-
A plug-and-play generative framework for multi-satellite precipitation estimation
PRISMA introduces a plug-and-play latent generative model that improves multi-sensor precipitation estimates by learning an unconditional prior from IMERG data and constraining it with independent sensor-specific branches.
-
InsightTok: Improving Text and Face Fidelity in Discrete Tokenization for Autoregressive Image Generation
InsightTok improves text and face fidelity in discrete image tokenization via content-aware perceptual losses, with gains transferring to autoregressive generation.
-
RoboEvolve: Co-Evolving Planner-Simulator for Robotic Manipulation with Limited Data
A co-evolutionary VLM-VGM loop on 500 unlabeled images raises planner success by 30 points and simulator success by 48 percent while beating fully supervised baselines.
-
Qwen-Image-VAE-2.0 Technical Report
Qwen-Image-VAE-2.0 achieves state-of-the-art high-compression image reconstruction and superior diffusability for diffusion models, with a new text-rich document benchmark.
-
VISOR: A Vision-Language Model-based Test Oracle for Testing Robots
VISOR is a VLM-based automated test oracle that evaluates robot task correctness and quality from videos while reporting its own uncertainty, tested on GPT and Gemini across four tasks and over 1000 videos with Gemini showing higher recall and GPT higher precision but low uncertainty-correctness tie
-
SceneFactory: GPU-Accelerated Multi-Agent Driving Simulation with Physics-Based Vehicle Dynamics
SceneFactory delivers a batched GPU platform for physics-based multi-agent autonomous driving simulation that achieves 127x higher throughput than non-vectorized PhysX while supporting articulated dynamics and road-condition friction.