Qwen it detect machine-generated text?
This work addresses the challenge of identifying AI-generated content for applications in content moderation and authenticity verification, representing an incremental improvement in a competitive benchmark.
The paper tackled the problem of detecting machine-generated text in a multilingual binary classification task, achieving first place in F1 Micro (0.8333) and second place in F1 Macro (0.8301) among 36 teams.
This paper describes the approach of the Unibuc - NLP team in tackling the Coling 2025 GenAI Workshop, Task 1: Binary Multilingual Machine-Generated Text Detection. We explored both masked language models and causal models. For Subtask A, our best model achieved first-place out of 36 teams when looking at F1 Micro (Auxiliary Score) of 0.8333, and second-place when looking at F1 Macro (Main Score) of 0.8301