pdf4llm.to_markdown("input.pdf") # get markdown for all pages splitter = MarkdownTextSplitter(chunk_size=40, chunk_overlap=0) splitter.create_documents([md_text]) For more see 5 Levels of Text Splitting