{
  "title": "Initialize from VLM weights and train on robot data",
  "caption": "The author made this diagram from Physical Intelligence’s π0 source figure. Internet-scale vision-language pretraining supplies the backbone. Robot training uses Open X-Embodiment and PI robot data. After this training, the policy can receive prompts or be fine-tuned for tasks. The diagram does not show robot pictures or task icons, and it is not a layer-exact computation graph. The playback is an example, and animation time is not measured runtime.",
  "alt": "Training flow: visual-language foundation weights combine with Open X-Embodiment and PI robot data to train a generalist π0 policy.",
  "sources": [
    "https://www.pi.website/blog/pi0"
  ],
  "limitations": [
    "The playback is an example. Playback time is not measured runtime.",
    "The diagram does not show a robot rollout or performance result.",
    "Dataset arrows show training inputs. They make no claim that all examples have all targets."
  ]
}