Skip to content

Latest commit

 

History

History
43 lines (30 loc) · 1.58 KB

File metadata and controls

43 lines (30 loc) · 1.58 KB

Concept

Proof of Autoresearch is built around one asymmetry:

Discovery is expensive. Verification is cheap.

Finding a useful training-recipe improvement may require many failed GPU runs and good hypothesis generation. Once a candidate improvement exists, a verifier can replay the same code change under controlled conditions and check whether the metric really improved.

The Loop

  1. A miner starts from a frozen seed training task.
  2. An LLM proposes a code diff against the allowed search surface.
  3. The miner runs the modified training job and records metrics.
  4. The miner bundles evidence: diff, config, environment, logs, and metrics.
  5. A validator fetches the evidence and replays the experiment.
  6. A paired bootstrap test rejects likely noise.
  7. In a live subnet, validator weights would reward verified improvements.

Why Bittensor

Bittensor already supplies a miner/validator incentive network. Proof of Autoresearch explores whether that network can score reproducible ML research work instead of benchmark responses, prompts, or arbitrary compute.

Why ARC

AutoResearch-Chain supplies the verification vocabulary:

  • content-addressed evidence bundles
  • frozen and searchable surfaces
  • commit-reveal submissions
  • replayable research claims

This repository combines those pieces into an ARC-Bittensor proof of concept.

Current Boundary

This is not a live subnet and not a profitability claim. The implementation has strong local tests and a mocked smoke path, but real IPFS, Docker replay, multi-machine operation, and testnet weight-setting remain the next milestones.