The document summarizes a study that benchmarked supervised learning models for emotion analysis on sentence-level text data. It evaluated one deep learning model (BiLSTM) and one machine learning model (FastText) on two datasets with six emotion categories. The results showed that while requiring significantly more training time, the BiLSTM model consistently outperformed the FastText models in terms of precision, recall, and F1-score. Text preprocessing was also found to improve the performance of both models.