· eval-pipeline

Upload dataset for eval

Drop a vendor-delivered dataset to run it through eval-pipeline (prep → exact match + LLM judge against gpt-5.5). Files land at raywardevalsrc/<vendor>/<asset_type>/<filename>; the runner picks them up automatically and writes a report you can browse.

Drag a file here, or click to browse
CSV · JSON · PDF · ZIP · uploads stream directly to Azure Storage