pubmed_clean for llm training
I’m interested in using this dataset to train a model, but I want to make sure this aligns with the licensing terms. I have some questions:
- Does the license permit using the dataset to train a model for research?
- Are there any attribution or redistribution requirements I should be aware of if I publish or share the trained model?
- Do you have any additional guidelines or restrictions on dataset usage beyond what’s in the license?
Thanks in advance!
Hi,
The dataset contains data from pubmed commercially available dataset. I believe it's called PMC Open access subset.
So yes, it's free to use.. that was the purpose so that I don't have to recreate it from scratch.
No additional requirements.
Just FYI,
If using for academic purpose, you can use non commercial files as well which are not included in this dataset.
Hi,
Thanks for this helpful information.
I am a bit confused though because I found a dataset here (https://huggingface.co/datasets/pmc/open_access) of open access articles and it says it contains approximately 3.4 million articles. However, your dataset contains 4.36 million rows. I am wondering how the difference is so big if your dataset is a subset of the open access dataset itself?
Thanks again,
Hmm.. that's strange, anyways, I will put the link from where the data has been taken: (https://pmc.ncbi.nlm.nih.gov/tools/pmcaws/), inside the oa_comm folder, there is a file going by oa_comm.filelist.csv, if you check the row size in there, it is close to 4.36 million rows. So I am not sure what "3.7 million articles" should translate to in terms of data. If you find something which is different from this information or anything in the dataset which is not under commercial terms, I am happy to refactor it. Apologies for the late reply.
Thanks..