The REALM checkpoint pretrained with CC-News as target corpus and Wikipedia as knowledge corpus, converted from the TF checkpoint provided by Google Language.
The original paper, code, and checkpoints can be found here.
from transformers import RealmEmbedder embedder = RealmEmbedder.from_pretrained("qqaatw/realm-cc-news-pretrained-embedder")
- Downloads last month
Unable to determine this model’s pipeline type. Check the docs .