mGPT-1.3B-ossetian / README.md
ai-forever's picture
Upload README.md with huggingface_hub
c828a68
metadata
language:
  - os
  - en
  - ru
license: mit
tags:
  - gpt3
  - transformers
  - mgpt

๐Ÿ† Ossetian mGPT 1.3B

Language model for Ossetian. Model has 1.3B parameters as you can guess from it's name.

Ossetian belongs to Indo-European language family. It's a very expressive language with approximately 0.5 million speakers. Here are some facts about it:

  1. It is spoken in the Caucasus region, predominantly in South and North Ossetia.
  2. Ossetian is descended from the ancient Scythian language.
  3. It has two main dialects: Iron and Digor.

Technical details

It's one of the models derived from the base mGPT-XL (1.3B) model (see the list below) which was originally trained on the 61 languages from 25 language families using Wikipedia and C4 corpus.

We've found additional data for 23 languages most of which are considered as minor and decided to further tune the base model. Ossetian mGPT 1.3B was trained for another 200 steps with batch_size=4 and context window of 2048 tokens on 1 A100.

Final perplexity for this model on validation is 18.7.

Chart of the training loss and perplexity:

Other mGPT-1.3B models

Feedback

If you'll found a bug of have additional data to train model on your language โ€” please, give us feedback.

Model will be improved over time. Stay tuned!