LLM-generated text detection on MELD-eval Claude-Haiku-4.5
100TPR @ 1% FPRMELD
Evaluation results
| Method | Links | |
|---|---|---|
| MELDEvaluation Protocol=zero-shot2026.05 | 100 | |
| ModernBERT-DetectEvaluation Protocol=zero-shot2026.05 | 98.2 | |
| RepreGuardEvaluation Protocol=zero-shot2026.05 | 41.2 | |
| Fast-DetectGPTEvaluation Protocol=zero-shot2026.05 | 14.3 | |
| RADAREvaluation Protocol=zero-shot2026.05 | 2.2 | |
| RoBERTa-Large (OpenAI)Evaluation Protocol=zero-shot2026.05 | 1.8 | |
| GLTREvaluation Protocol=zero-shot2026.05 | 1.3 | |
| BinocularsEvaluation Protocol=zero-shot2026.05 | 1.1 | |
| Human-as-OutlierEvaluation Protocol=zero-shot2026.05 | 0.8 | |
| RoBERTa-ChatGPTEvaluation Protocol=zero-shot2026.05 | 0.5 |