load_dataset save_to_disk load_from_disk error

[brisker]

Describe the bug

datasets version： 2.10.1
I load_dataset and save_to_disk sucessfully on windows10( and I load_from_disk(/LLM/data/wiki) succcesfully on windows10), and I copy the dataset /LLM/data/wiki
into a ubuntu system, but when I load_from_disk(/LLM/data/wiki) on ubuntu, something weird happens:

load_from_disk('/LLM/data/wiki')
  File "/usr/local/miniconda3/lib/python3.8/site-packages/datasets/load.py", line 1874, in load_from_disk
    return DatasetDict.load_from_disk(dataset_path, keep_in_memory=keep_in_memory, storage_options=storage_options)
  File "/usr/local/miniconda3/lib/python3.8/site-packages/datasets/dataset_dict.py", line 1309, in load_from_disk
    dataset_dict[k] = Dataset.load_from_disk(
  File "/usr/local/miniconda3/lib/python3.8/site-packages/datasets/arrow_dataset.py", line 1543, in load_from_disk
    fs_token_paths = fsspec.get_fs_token_paths(dataset_path, storage_options=storage_options)
  File "/usr/local/miniconda3/lib/python3.8/site-packages/fsspec/core.py", line 610, in get_fs_token_paths
    chain = _un_chain(urlpath0, storage_options or {})
  File "/usr/local/miniconda3/lib/python3.8/site-packages/fsspec/core.py", line 325, in _un_chain
    cls = get_filesystem_class(protocol)
  File "/usr/local/miniconda3/lib/python3.8/site-packages/fsspec/registry.py", line 232, in get_filesystem_class
    raise ValueError(f"Protocol not known: {protocol}")
ValueError: Protocol not known: /LLM/data/wiki

It seems that something went wrong on the arrow file?
How can I solve this , since currently I can not save_to_disk on ubuntu system

Steps to reproduce the bug

datasets version： 2.10.1

Expected behavior

datasets version： 2.10.1

Environment info

datasets version： 2.10.1

[brisker]

solved.
fsspec version problem

[beyondguo]

I'm using the latest datasets and fsspec , but still got this error!

datasets : Version: 2.13.0

fsspec Version: 2023.10.0

File "/home/guoby/app/Anaconda3-2021.05/envs/news/lib/python3.8/site-packages/datasets/load.py", line 1892, in load_from_disk
    return DatasetDict.load_from_disk(dataset_path, keep_in_memory=keep_in_memory, storage_options=storage_options)
  File "/home/guoby/app/Anaconda3-2021.05/envs/news/lib/python3.8/site-packages/datasets/dataset_dict.py", line 1371, in load_from_disk
    dataset_dict[k] = Dataset.load_from_disk(
  File "/home/guoby/app/Anaconda3-2021.05/envs/news/lib/python3.8/site-packages/datasets/arrow_dataset.py", line 1639, in load_from_disk
    fs_token_paths = fsspec.get_fs_token_paths(dataset_path, storage_options=storage_options)
  File "/home/guoby/app/Anaconda3-2021.05/envs/news/lib/python3.8/site-packages/fsspec/core.py", line 610, in get_fs_token_paths
    chain = _un_chain(urlpath0, storage_options or {})
  File "/home/guoby/app/Anaconda3-2021.05/envs/news/lib/python3.8/site-packages/fsspec/core.py", line 325, in _un_chain
    cls = get_filesystem_class(protocol)
  File "/home/guoby/app/Anaconda3-2021.05/envs/news/lib/python3.8/site-packages/fsspec/registry.py", line 232, in get_filesystem_class
    raise ValueError(f"Protocol not known: {protocol}")

[AlisonWen]

These two versions work.

[robinsonmhj]

datasets==2.10.1 and fsspec==2023.6.0 also works for me.

[1412690667]

确实