Loading dataset details…
harbor run -d harbor-index/harbor-index-1.0harbor run -d harbor-index/harbor-index-1.0Harbor Index is an 82-task benchmark for agentic evaluation. It was distilled from more than 6,000 candidate tasks through repeated model runs, automated broken-task identification, human audits, and reward hacking supervision.
harbor run -d harbor-index/harbor-index-1.0Harbor Index is an 82-task benchmark for agentic evaluation. It was distilled from more than 6,000 candidate tasks through repeated model runs, automated broken-task identification, human audits, and reward hacking supervision.
harbor run -d harbor-index/harbor-index-1.0