This study investigates the efficacy of prompting techniques in enhancing intent classification accuracy within scientific citations, drawing inspiration from their successful application in datasets like GSM8K and seminal research such as CitePrompt [1]. Despite the demonstrated potential of prompting in achieving state-of-the-art results across diverse domains, there remains a notable gap in the comprehensive exploration of prompting techniques themselves. Motivated by the insights from CitePrompt and driven by a desire to delve deeper into the nuances of prompting methodologies, we undertake an approach that involves removing the verbaliser and transformers, focusing solely on the core prompting techniques.

1] Avishek Lahiri, Debarshi Kumar Sanyal, Imon Mukherjee 2023. CitePrompt: Using Prompts to Identify Citation Intent in Scientific Papers