Datasets:
Languages:
English
Multilinguality:
multilingual
Size Categories:
10M<n<100M
Language Creators:
crowdsourced
Annotations Creators:
no-annotation
Source Datasets:
original
ArXiv:
License:
What exactly mean for `no_index` subset?
#6
by
seonglae
- opened
According to readmd, they have same columns
The no_index subset doesn't download the DPR embeddings alongside the dataset.
Isn't it no_embedding?
Oh, you are correct. This still downloads the embeddings but doesn't build a FAISS index on them.
Oh that's why the pulling process of the exact subset is quite slow. Thanks for the information
seonglae
changed discussion status to
closed