SINAI Participation in SimpleText Task 2 at CLEF 2024: Zero-shot Prompting on GPT-4-Turbo for Lexical Complexity Prediction

Abstract:

In this article, we present our participation in Tasks 2.1 and 2.2 of the SimpleText track of CLEF 2024. Our work focused on the implementation of zero-shot learning using the GPT-4 Turbo autoregressive model. To this end, we develop and evaluate various cues to optimize the model's ability to predict lexical complexity. The results of our experiments indicated that GPT-4 Turbo can perform this task with remarkably robust performance, demonstrating its potential to assess language complexity effectively without the need for additional training.

Año de publicación:

2024

Keywords:

• Auto-regressive models
• GPT-4 Turbo
• Lexical Complexity Prediction
• Prompting
• Zero-Shot learning

Fuente:

scopus