YYLO Benchmark compares models, harnesses and configurations: reviewed cases produce independent attempts, and retained outputs can be evaluated with different checks or judges later. It does not choose a winner or combine judge opinions with test results. Cases are prepared once from historical Ledger tasks, supplied coding prompts, or workflows, and yylo-benchmark runs alongside the YYLO CLI and YYLO Ledger packages.
โ
RETURN
YYLO Benchmark
A thin trusted-host experiment runner for historical Ledger tasks, supplied coding prompts, and workflows.