chunked/nltk_chunking/model_doc_patchtsmixer/chunk_14.txt

It also outperforms the latest strong benchmarks of Patch-Transformer models (by 1-2%) with a significant reduction in memory and runtime (2-3X).