This paper provides different approaches for a binary sentiment classification on a small training dataset. LLMs that provided state-of-the-art results in sentiment analysis and similar domains are being used, such as BERT, RoBERTa and XLNet.