Call for Contributions

Contribute closed-loop experimental tasks from your cloud lab to Autonomous Science Bench.

Autonomous Science Bench 0.1Deadline TBD
  1. 01Propose
    Task proposal form (coming soon)
  2. 02Implement
    Placeholder: task specification and experiment interface.
  3. 03Verify
    Placeholder: review, reference method, and independent verifier.

About

PlaceholderWhat Autonomous Science Bench is, the network of programmable cloud lab nodes behind it, and how contributed tasks feed back into better scientific agents.
Cloud lab node network
PlaceholderHow domains are organized and which fields are open to new tasks.
  • Biology
    FieldFieldField
  • Chemistry
    FieldFieldField
  • Materials
    FieldFieldField
  • Semiconductors
    FieldFieldField
  • Other sciences
    FieldFieldField

Contribute

What we're looking for

PlaceholderThe kinds of experimental workflows that make good tasks.
  • Closed-loop

    Placeholder: the task rewards learning from each round's results.

  • Experimentally verified

    Placeholder: success is checked against measurements, not a proxy.

  • Fully budgeted

    Placeholder: rounds, batch sizes, costs, and compute are fixed up front.

What each node supplies

  • 01Task specificationPlaceholder
  • 02DataPlaceholder
  • 03Experiment interfacePlaceholder
  • 04Reference methodPlaceholder
  • 05Independent verifierPlaceholder

Why contribute?

PlaceholderWhat nodes and contributors gain: influence on frontier AI, evaluation experience, and authorship.

How to contribute

PlaceholderThe propose → implement → verify process in detail, with review stages and timelines.

Credit and responsibilities

PlaceholderAuthorship policy and what contributors maintain after their task is merged.

Resources

PlaceholderLinks to the task format, contributing guide, example tasks, and the proposal board.