Twitter sentiment analysis with BeRT (Classification F1 score: 0.71 -8th-, Regression RMSE: 0.20 -3rd-)
We selected to use the BERT model for our sentiment analysis project. BERT, which stands for Bidirectional Encoder Representations from Transformers, is a leading-edge language model known for its strong grasp of context and textual nuances, making it an ideal fit for sentiment analysis.
We ended up being 8th in classification in the class and 3rd in regression.