One of our papers was accepted to EMNLP 2025 Findings!
Evaluating Text Generation Quality Using Spectral Distances of Surprisal
The paper introduces spectral distances of surprisal as a new perspective for evaluating the quality of generated text.
← Back to news
A Paper Accepted to EMNLP 2025 Findings
Our work on evaluating text generation with spectral distances of surprisal was accepted to EMNLP 2025 Findings.