A new benchmark with synthetic PDFs and LLM semantic judging reveals wide performance gaps among PDF parsers on math formulas and validates the judge via human ratings at r=0.78.
https://blog.google/products/gemini/gemini-3/ (2025), accessed: 2025-12-01
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Benchmarking Document Parsers on Mathematical Formula Extraction from PDFs
A new benchmark with synthetic PDFs and LLM semantic judging reveals wide performance gaps among PDF parsers on math formulas and validates the judge via human ratings at r=0.78.