chunked/nltk_chunking/model_doc_deplot/chunk_3.txt

Prior state-of-the-art (SOTA) models require at least tens of thousands of training examples and their reasoning capabilities are still much limited, especially on complex human-written queries.