DeRA: Decoupled Representation Alignment for Video Tokenization

arXiv 2025 29.6 method

TLDR

DeRA is a 1D video tokenizer that decouples spatial and temporal representation learning, aligning with vision foundation models to improve video generation efficiency and performance.

Reasoning

The paper presents a novel video tokenizer with a decoupled architecture and a gradient conflict resolution module, showing strong empirical results on video generation benchmarks. However, the abstract does not position the work as a world model, so relevance to world model keywords is limited to indirect connections via video generation.

Read-first score

Read-first score 29.6, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 4.

Recency 8%

86.7

Uses a gentle age decay so recent papers surface without erasing older foundations. 2025

Methodology quality 25%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=experiment,metric,result

Reproducibility 25%

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

Topical relevance 42%

5.7

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

Field roles

Frontier

Rank sensitivity

Stability: volatile; rank range: 118.

Keyword Scores

video world model

world model

world dynamics prediction

world simulator

generative world model

interactive world model

model-based reinforcement learning world model

Resources

Paper