Mark Gales

This article explores how text elements influence classification tasks in NLP.

This framework improves text evaluation efficiency and accuracy using Large Language Models.

CrossCheckGPT provides a new way to evaluate model reliability and accuracy.

A universal audio clip can mute advanced ASR models like Whisper.

Research reveals risks in multi-task speech models like Whisper.

A new method enhances text evaluation by using soft probabilities for better accuracy.