|
--- |
|
language: |
|
- ru |
|
- myv |
|
tags: |
|
- erzya |
|
- mordovian |
|
- fill-mask |
|
- pretraining |
|
- embeddings |
|
- masked-lm |
|
- feature-extraction |
|
- sentence-similarity |
|
license: cc-by-sa-4.0 |
|
datasets: |
|
- slone/myv_ru_2022 |
|
--- |
|
|
|
This is a version of [LaBSE-en-ru-myv-v1](https://huggingface.co/slone/LaBSE-en-ru-myv-v1), fine-tuned for about 150K steps |
|
on the [myv_ru_2022](https://huggingface.co/datasets/slone/myv_ru_2022) dataset, in |
|
[this notebook](https://colab.research.google.com/drive/1SxeraKZS6KYKobzVNNyIQZa4WnhpJ_nb?usp=sharing). |
|
|
|
It demonstrates slighly better results than the v1 model, both on bitext mining and on the MLM task. |
|
|
