meta harness
Meta-Harness’s frontier rule, as a HarnessGrad method.
The method returns one edited harness per round. It never sees a score for the harness it is producing — the platform measures every candidate on the same study set and records the result in curve.jsonl.
Measured results
Section titled “Measured results”Scores, rounds and per-split comparisons are on the ALFWorld benchmark page — one table for every method, so the same number cannot say two different things in two places.
Record
Section titled “Record”This method’s rounds on the Hub — each round’s harness, the edit that produced it, and its per-task trajectories: open the directory.
Deviations from the method it ports
Section titled “Deviations from the method it ports”5 deliberate difference(s), recorded in methods/meta_harness/method.json:
- A different experiment on a different model, not a reproduction.
- History depth is the newest 12 states, not every prior candidate.
- The gate cannot smoke-test.
- A bounded two-stage batch replaces the upstream agent-written candidate manifest.
- Design is an endpoint JSON call, rather than the upstream trajectory-reading subagent.
methods/meta_harness/run.py and its rule test in tests/quality/. python3 tools/method_lint.py prints one line per method.