olmOCR-Bench
An OCR and document parsing evaluation framework released by Allen AI. It replaces traditional statistical metrics such as CER and BLEU by decomposing fuzzy similarity measures into binary Boolean checks targeting specific page content elements, enabling more precise assessment of document parsing quality.
Timeline (last 90 days)
OXR officially claims to rank first simultaneously on three mainstream document parsing benchmarks: olmOCR-Bench, OmniDocBench v1.6, and DocParsingBench.
On olmOCR-Bench, OXR-1.0 scores 87.84, and the second-place Infinity-Parser2-Pro scores 87.60, a gap of only 0.24 points.
olmOCR-Bench is a document OCR evaluation benchmark launched by AI2 (Allen Institute for AI), built on real PDFs and covering diverse layouts such as tables, formulas, handwriting, and multiple columns, with scoring that assesses both text accuracy and reading order.
olmOCR-Bench is an OCR and document parsing evaluation framework released by Allen AI, whose core design philosophy is to decompose the overall similarity between output and reference text into binary Boolean checks targeting specific page content.
All Facts (4)
OXR officially claims to rank first simultaneously on three mainstream document parsing benchmarks: olmOCR-Bench, OmniDocBench v1.6, and DocParsingBench.
50%UnverifiedOn olmOCR-Bench, OXR-1.0 scores 87.84, and the second-place Infinity-Parser2-Pro scores 87.60, a gap of only 0.24 points.
50%UnverifiedolmOCR-Bench is a document OCR evaluation benchmark launched by AI2 (Allen Institute for AI), built on real PDFs and covering diverse layouts such as tables, formulas, handwriting, and multiple columns, with scoring that assesses both text accuracy and reading order.
50%UnverifiedolmOCR-Bench is an OCR and document parsing evaluation framework released by Allen AI, whose core design philosophy is to decompose the overall similarity between output and reference text into binary Boolean checks targeting specific page content.
50%