Is the full data ready?

#6
by shizhediao2 - opened

Hi,

Thank you so much for such a great contribution! I was wondering if the dataset is ready? Since I found there are some empty files (e.g., https://huggingface.co/datasets/LLM360/TxT360/blob/main/data/pubmed_central/1001-inf/0_0.jsonl). Are there any timelines?

Thank you!

Hi, @shizhediao2 ,

I believe the data is ready. It is very likely that we don't have data fitting that criterion (the file name means a doc duplicates for 1001 or more times, which is unlikely for any paper).

@nikhilranjan , I believe that's the case right? We probably should also remove empty files like this to reduce confusion.

LLM360 org

Hi @shizhediao2 ,

Thanks you for your interest in our dataset and for bringing this to our attention.

Yes, I also believe the full data is ready.

From my side, the example file you mentioned seems not empty but contains one line, as shown in the screenshot below:
image.png

This particular file is under the folder "1001-inf". The folder name "1001-inf" refers to the number of duplicates for the given document, indicating that only one document has such a high number of duplicates in the pubmed_central subset.

p.s., we are actively working to add details of our folder structure to minimize confusion for users. This update should be available shortly.

Could you please let us know if you have found any other empty files? This will help us ensure there are no problems with our upload process.

Thank you again for your support!

Sign up or log in to comment