This model is mainly focussed on extracting the sentiment on tweets regarding bitcoin. The model was trained on manually on labeled data with rubrix (https://www.rubrix.ml/). The training set approximately contained 500 samples and 500 test samples. The cardiffnlp/twitter-xlm-roberta-base-sentiment (https://huggingface.co/cardiffnlp/twitter-xlm-roberta-base-sentiment) was used as weak classifier and also as base-model for finetuning.