Hi, everyone.
After some data gathering, I managed to do ULMFIT in the Estonian language. I’ve used oscar data for training the language model. For fine-tuning, I couldn’t find a suitable dataset do I crawled some news websites and got around 2700 news articles in three categories. Though it’s a tiny dataset, I gathered this amount of data for the sake of demonstration. So it will be really great if Estonian folks try this model and fine-tune it on their own dataset 
.
You can check out the procedures in this blog post for more details:
2 Likes