Upload a PDF document to run LayoutLMv3 inference (BIO-Tagging) followed by post-processing to generate a clean, structured output (JSON). Note: The post-processing logic needs to be finalized with your custom code.