Performance benchmarks: read → pseudo-translate → write across a corpus of real-world files
Loading benchmark data...