Goal is to create a strong language generation model for Mongolian Since initial code and data is pretty much written by @patrickvonplaten and other huggingface members, it should not be so hard to get the first sense.

Randomly initialized GPT2 model

We can use OSCAR which is available through datasets

A causal language modeling script for Flax is available here 1. It can be used pretty much without any required code changes. If there is time left, I’d love to try some private crawling and integrate it datasets.

Understandable Mongolian text generation model

Lack of data → OSCAR Mongolian is just 2.2G. Maybe we need to research ways to acquire more data with this.