Drop a folder's worth of resumes. Each one is run through every configured extractor, then an independent model is shown the rendered pages and both extractions — blinded and in random order — and asked which is more accurate.
started 09:34:09 · single document