Ir al contenido

All commands

chimera transfer-gate

Promote a learned change only if it helps its tuned slice AND doesn't regress a holdout. Guards against *negative transfer* — a GEPA prompt / ACE delta / distilled skill that raises the pass rate on the tasks it was tuned against but REGRESSES on other tasks sharing the capability. Feed the tuned slice's paired pass/fail (baseline vs candidate) and, ideally, a disjoint same-capability holdout's; the verdict is PROMOTE / BLOCK with the paired evidence (exit 1 on BLOCK, for CI). Without a holdout it promotes on the tuned gain alone but flags that transfer was NOT measured.

Arguments

  • TUNED_BASELINEstrrequired

    JSON pass/fail of the baseline on the TUNED slice (list of bools, or {task: bool}).

  • TUNED_TREATMENTstrrequired

    JSON pass/fail of the candidate on the TUNED slice (aligned, same order).

Options

  • --holdout-baselinestr

    JSON pass/fail of the baseline on a DISJOINT same-capability holdout.

  • --holdout-treatmentstr

    JSON pass/fail of the candidate on the holdout (aligned).

  • --require-significantboolean

    Require the tuned gain's paired CI to exclude 0, not just Δ>0.

  • --tolfloatdefault: 0.0

    Max tolerated pass-rate drop on the holdout before promotion is blocked.