chunked/nltk_chunking/model_doc_poolformer/chunk_10.txt

Based on the extensive experiments, we argue that MetaFormer is the key player in achieving superior results for recent transformer and MLP-like models on vision tasks.