This article investigates the performance of models on social media posts by examining different Transformer architectures. It evaluates the effectiveness of fine-tuning and prompt-based methods, aiming to identify best practices and trade-offs through performance comparisons across various models.