Model for predicting geolocation by texts. Tweets from South America were used for training.
For testing the model the test file (such as the example file -- test.csv) is necessary.
To finetune model the source model, tokenized texts and labels are necessary.