Transforming Next-Token Prediction into Classification with LLMs

From tokens to labels: Performing classification with large language models