Divot: Diffusion Powers Video Tokenizer for Comprehension and Generation

arXiv 2024 35.6 method

TLDR

Divot introduces a diffusion-powered video tokenizer that unifies video comprehension and generation in LLMs, achieving competitive benchmark performance.

Reasoning

The paper presents a novel tokenizer leveraging diffusion for self-supervised video representation learning and a diffusion-based de-tokenizer, with strong empirical results on video benchmarks. However, the abstract does not explicitly connect the method to world models or dynamics prediction, limiting relevance to those keywords.

Read-first score

Read-first score 35.6, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 9.

Recency 8%

75.1

Uses a gentle age decay so recent papers surface without erasing older foundations. 2024

Methodology quality 25%

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

Reproducibility 25%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=code,github

Topical relevance 42%

12.9

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

Candidate

Rank sensitivity

Stability: volatile; rank range: 79.

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