AdaJEPA: An Adaptive Latent World Model

arXiv 2026 36.3 method

TLDR

AdaJEPA adapts a latent world model at test time via self-supervised updates within MPC, improving planning under distribution shift.

Reasoning

The paper introduces a novel test-time adaptation mechanism for latent world models, addressing a key limitation of frozen models. Its strength lies in the closed-loop self-supervised update that requires no extra demonstrations. However, the abstract lacks details on task diversity and does not mention real-world validation, limiting generalizability claims.

Read-first score

Read-first score 36.3, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 43.

Recency 6%

100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Topical relevance 29%

61.4

Uses existing LLM keyword relevance scores normalized to 0-100. world model,world simulator,generative world model,interactive world model,video world model,world dynamics prediction,model-based reinforcement learning world model

Methodology quality 18%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=none

Reproducibility 18%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=none

Citation impact 18%

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. cited_by_count=0

Citation velocity 12%

Citation velocity estimates citations per publication-year to reduce old-paper bias. velocity=0.00

Field roles

Frontier

Rank sensitivity

Stability: volatile; rank range: 221.

Keyword Scores

world model

world dynamics prediction

model-based reinforcement learning world model

interactive world model

world simulator

generative world model

video world model

Deep Analysis

Innovations

Test-time adaptation of a latent world model within the closed loop of model predictive control (MPC)
Self-supervised adaptation using observed next-state transitions as a signal, without additional expert demonstrations
Continuous recalibration of the world model with as few as one gradient step per MPC replanning step

Methodology

AdaJEPA is a latent world model that, after training, performs MPC by planning action chunks, executing the first chunk, then using the observed next-state transition as a self-supervised adaptation signal to update the model (one gradient step) before replanning, all in a closed loop.

Key Results

Across a range of goal-reaching tasks, AdaJEPA substantially improves planning success with as few as one gradient step per MPC replanning step.

Resources

Paper