{
  "title": "Context adaptation or policy-weight updates",
  "caption": "Figure 10. GEN-1.5 uses a 3–12 second sensorimotor demonstration in a 30-second rolling context, and the developer’s report gives a 100 Hz rate for action trajectories. The developer’s report gives 59% ±10% without gradient steps for ten simple short-horizon tasks. The 83% ±9% result uses approximately five minutes of task data and ten gradient steps. The rates are not comparable across benchmarks. The author made this diagram. Context updates do not change weights, and tuning is a separate condition.",
  "alt": "Adaptation flow supplies demonstrations for a rolling context with fixed weights, and a separate dataset and gradient tuning can change weights.",
  "sources": [
    {
      "path": "content/blog/2026-09-08_robot-use-agents/post.md",
      "description": "Article explanation and retained source citations"
    }
  ],
  "limitations": [
    "The execution is an example. Playback duration is not measured runtime.",
    "The developer’s report gives 59% ±10% for ten selected short-horizon tasks without gradients. The 83% ±9% result is from a separate tuning condition. This playback measures no online interpolation or causal performance gain."
  ]
}