mlsanityData to 7 Oct 2026

DeepJEPA: Scaling World Models from Within

arXiv 2610.00368 ↗First posted 30 Sep 2026cs.RO cs.AIcited by 0

On the map: World models → Latent predictive world models

Abstract (arXiv)

World-model planners typically scale outward by rolling farther, sampling more trajectories, or optimizing longer, while assigning the same computation to every imagined transition. We show that making every transition uniformly deeper wastes computation and can degrade planning because useful refinement is concentrated at a small set of decision-critical events. We introduce DeepJEPA, a weight-tied joint-embedding predictive world model that treats transition depth as an inner test-time scaling axis and learns when another recurrent update is worth computing for each candidate and rollout step. Across five visual-control settings, DeepJEPA improves or matches the strongest fixed-depth planner while averaging only 1.00-1.26 updates per transition. Its additional computation concentrates at contact onset and sustained object interaction, where latent corrections can change which candidates enter the planner's elite set and which action is selected. Representation probes further show that improved planning does not require uniformly better object-state decodability. DeepJEPA therefore reframes world-model scaling as a problem of allocating internal computation where it can change the planner's decision: think deeper at decision-critical transitions instead of making every rollout uniformly deeper or longer.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Latent predictive world models, one of 11 directions in World models. Builds on 32 of its 62 references that are in the map. First posted on arXiv 30 Sep 2026.

Builds on

32 of 62 references.

Closest ancestors · 4

Same direction, newest first

  1. 2026Hierarchical Planning with Latent World Models3 Apr
  2. 2026LeWorldModel: Stable End-to-End Joint-Embedding Predictive Architecture from Pixels13 Mar
  3. 2026Learning Invariant Visual Representations for Planning with Joint-Embedding Predictive World Models20 Feb
  4. 2024Revisiting Feature Prediction for Learning Visual Representations from Video15 Feb

Other references · 28

  1. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,601
  2. 2023RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Controlcited by 2,993
  3. 2024OpenVLA: An Open-Source Vision-Language-Action Modelcited by 2,777
  4. 2022RT-1: Robotics Transformer for Real-World Control at Scalecited by 2,007
  5. 2018World Modelscited by 1,444
  6. 2019Mastering Atari, Go, Chess and Shogi by Planning with a Learned Modelcited by 1,327
  7. 2023Mastering Diverse Domains through World Modelscited by 1,193
  8. 2024Large Language Monkeys: Scaling Inference Compute with Repeated Samplingcited by 946
  9. 2019When to Trust Your Model: Model-Based Policy Optimizationcited by 753
  10. 2018Deep Reinforcement Learning in a Handful of Trials using Probabilistic Dynamics Modelscited by 748
  11. 2025V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planningcited by 738
  12. 2018Universal Transformerscited by 613
16 more
  1. 2016Adaptive Computation Time for Recurrent Neural Networkscited by 479
  2. 2026World Action Models are Zero-shot Policiescited by 411
  3. 2025WorldVLA: Towards Autoregressive Action World Modelcited by 351
  4. 2026Fast-WAM: Do World Action Models Need Test-time Future Imagination?cited by 311
  5. 2016Deep Visual Foresight for Planning Robot Motioncited by 276
  6. 2025LeJEPA: Provable and Scalable Self-Supervised Learning Without the Heuristicscited by 227
  7. 2021PonderNet: Learning to Pondercited by 156
  8. 2026VLA-JEPA: Enhancing Vision-Language-Action Model with Latent World Modelcited by 104
  9. 2025World Models for Learning Dexterous Hand-Object Interactions from Human Videoscited by 26
  10. 2026Light-WAM: Efficient World Action Models with State-Fusion Action Decodingcited by 19
  11. 2025Adapting World Models with Latent-State Dynamics Residualscited by 8
  12. 2026Is Your Driving World Model an All-Around Player?cited by 5
  13. 2026Data Pyramid for Embodied Manipulation: A Surveycited by 4
  14. 2014Selecting Computations: Theory and Applicationscited by 3
  15. 2026Adapting in the Dark: Efficient and Stable Test-Time Adaptation for Black-Box Modelscited by 2
  16. 2026Not All Points Are Equal: Uncertainty-Aware 4D LiDAR Scene Synthesiscited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026The Planning Limits of Latent World Models
  2. 2026VisualPatchWorld: Code World Models as Latent Structured Representations for Planning
  3. 2020World Model as a Graph: Learning Latent Landmarks for Planning
  4. 2026JEPA-TTT: Persistent Test-Time Training of Latent World Models for Planning under Dynamics Shifts
  5. 2025World-in-World: World Models in a Closed-Loop World
  6. 2026LRC-JEPA: Disentangling Dynamics and Residual Context for Efficient World Models
  7. 2026AdaJEPA: An Adaptive Latent World Model
  8. 2026H-JEPA: End-to-End Learning of Hierarchical World Models for Visual Planning