Missing Specific Directories After Dataset Download
Hello, thank you very much for your contribution. I have downloaded the complete dataset, but I noticed that the directories/files highlighted in the image were not downloaded. When I tried to download these specific directories individually, it showed that there were no files available for download. Could you please help me understand what might be causing this issue?
Hi, @zhengxvc
Thanks for being interested in our dataset. I just check a few directory you mentioned, it seems that I can see the files, as shown in the screenshots, would you mind let us know which specific directory is empty for you?
https://huggingface.co/datasets/LLM360/TxT360/tree/main/data/europarl/1-1
Thank you for your reply. Under the "data" directory, all directories except for "arxiv" and "common-crawl" do not exist, yet it shows that the download is complete.
hmm. In my screenshot which I just took on the page, it seems the data is there. Would you mind if you can elaborate in what context you find these files "do not exist", it could be due to some incompatibility with data downloader you are using?
I downloaded the dataset using the following command:
"huggingface-cli download --repo-type dataset --resume-download LLM360/TxT360 --local-dir LLM360__TxT360 --local-dir-use-symlinks False --token {token}"
After the download was completed, I checked the results and found this situation.
i see. it might be that there's something of the dataset that is not compatible the huggingface downloader. @nikhilranjan maybe we can take a look? or do you have any insights?
at the moment, maybe you can try some alternative download methods?
Hi @zhengxvc , is it due to local cache issue ? I would suggest to download via git-lfs command and later read using pandas dataframe or jsonl libary before using huggingface libraries like datasets. We are trying to work on arrow compatibility so that it can be directly loaded as hf intended but still work in progress.
i see. it might be that there's something of the dataset that is not compatible the huggingface downloader. @nikhilranjan maybe we can take a look? or do you have any insights?
at the moment, maybe you can try some alternative download methods?
Thank you for the reply. It is very likely due to compatibility issues. I updated the version of Hugging Face, but it didn't resolve the problem. I will try other download methods.
Hi @zhengxvc , is it due to local cache issue ? I would suggest to download via git-lfs command and later read using pandas dataframe or jsonl libary before using huggingface libraries like datasets. We are trying to work on arrow compatibility so that it can be directly loaded as hf intended but still work in progress.
Hello, I have cleared the local cache and re-downloaded, but the issue still persists, so I believe the problem with the local cache can be ruled out. Okay, I will try the method you suggested. Thank you for your help.
Hi.@hunterhector @nikhilranjan , I have already resolved the issue. It was indeed a problem with the local cache. The files I cleaned up earlier were not the actual cache. After I deleted the cache and re-downloaded everything, it worked fine.
@zhengxvc can you please tell which method did you use to download the data I am getting some errors like the below one continuously
File "/usr/local/lib/python3.10/dist-packages/huggingface_hub/file_download.py", line 1599, in _raise_on_head_call_error
raise LocalEntryNotFoundError(
huggingface_hub.errors.LocalEntryNotFoundError: An error happened while trying to locate the file on the Hub and we cannot find the requested files in the local cache. Please check your connection and try again or make sure your Internet connection is on.
@zhengxvc can you please tell which method did you use to download the data I am getting some errors like the below one continuously
File "/usr/local/lib/python3.10/dist-packages/huggingface_hub/file_download.py", line 1599, in _raise_on_head_call_error
raise LocalEntryNotFoundError(
huggingface_hub.errors.LocalEntryNotFoundError: An error happened while trying to locate the file on the Hub and we cannot find the requested files in the local cache. Please check your connection and try again or make sure your Internet connection is on.
@sravan-iitb I am using the command "huggingface-cli download --repo-type dataset --resume-download LLM360/TxT360 --local-dir LLM360__TxT360 --local-dir-use-symlinks False --token {token}" to download the dataset. The local cache can be cleared by deleting the .cache file in the target folder.