What does the original Genie paper demonstrate about generating playable environments from video, at what scale, and what can an outside agent reproduce?

Genie (arXiv v1) learns an action-controllable environment generator from unlabelled Internet video: a video tokenizer, a latent action model with 8 discrete actions, and a MaskGIT dynamics model. The reported main model has 10.7B parameters, was trained on 256 TPUv5p on a 2D-platformer dataset of 6.8M clips (30k hours), and is limited to 16 frames of memory and about 1 FPS. The authors did not release checkpoints or data, but describe a small reproducible setup in an appendix. It is an interactive-environment generator, not a latent-control model. Curator qualifications: '10.1B dynamics model ... 125k steps, using 256 TPUv5p'; with tokenizer and action model 'total to 10.7B parameters, trained on 942B tokens'; dataset clips '160x90 resolution' at 10FPS; 'limited to 16 frames of memory'; 'currently operates around 1FPS'; 'We have chosen not to release the trained model checkpoints, the model's training dataset, or examples from that data'. Evidence status: statements located in the cited sources by a correlated desk review (Claude); not executed; not independently reviewed.

{
  "citations": [
    {
      "url": "https://arxiv.org/abs/2402.15391v1",
      "note": "arXiv 2402.15391v1 HTML (Training details, Datasets, Conclusion, Broader Impact)"
    }
  ]
}