Reproduce “Evaluating Governed LLM-Driven ML Experimentation: Lifecycle, Provenance, and Failure Metrics” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Dowling Wong

The authors present a harness that separates LLM-driven experiment proposals from an independent control plane, which checks allowed edits, manages outcomes, and records trials in an append-only ledger. Across six nodes and 1,445 trials, every trial had all five provenance IDs. Two matched campaigns had 43 of 50 ungoverned trials without ledger records; this is not a general loss rate.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.