Can't load dataset

#1
by elviswxy - opened

Hi there,

It seems that the dataset is outdated as I can't call dataset = load_dataset("sentence-transformers/msmarco-hard-negatives") to load the dataset.
My env:
datasets 2.15.0
transformers 4.35.2
How can I solve this issue? Thanks.

Sentence Transformers org

Hello!

I'm afraid that this dataset isn't yet set up nicely - it can't be loaded with load_dataset right now. You have to manually download the files or automatically download them using the full URL to the files.

  • Tom Aarsen

@tomaarsen Could you share with me any documents about how this dataset is generated? The MS MARCO passage retrieval dataset has 532,761 training pairs, however, in your msmarco-hard-negatives.jsonl.gz, there are 808730 queries which is much more than the # of queries in the official training dataset. Thanks.

Zhiyuan.

Sign up or log in to comment