chunked/nltk_chunking/model_doc_code_llama/chunk_19.txt

As mentioned above, the dtype of the storage weights is mostly irrelevant unless you are using torch_dtype="auto" when initializing a model using.