russian fill-mask pretraining embeddings masked-lm

!!! At the moment, the model is distilled, a version from one of the first checkpoints is available for download. We plan to post the full model in the next few days. !!!

This is a distilled HRBert model for an mlm task.

Sentence embeddings can be produced as follows:


# pip install transformers

from transformers import pipeline


fill_mask = pipeline(
    "fill-mask",
    model='RabotaRu/HRBert-mini',
    tokenizer='RabotaRu/HRBert-mini'
)

fill_mask('<mask> на склад')