Get clean data from tricky documents, powered by vision-language models ⚡ - View it on GitHub