chunked/nltk_chunking/model_doc_layoutlmv2/chunk_43.txt

This is not a problem, as these parameters are batch normalization
  statistics, which are going to have values when fine-tuning on a custom dataset.