AI-generated text detection on master (test)
99.6AccuracyBest LLM Finetune (Phi-4)
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Best LLM Finetune (Phi-4)number of parameters=14.7B2026.01 | 99.6 | 99.6 | 99.8 | 99.3 | 99.6 | |
| Open-source SOTAnumber of parameters=430M2026.01 | 88 | 87.2 | 84 | 96.2 | 89.7 | |
| Per LLM Family Ensemblenumber of parameters=69B2026.01 | 80.6 | 79.2 | 75.1 | 96.1 | 84.3 | |
| Per LLM Ensemblenumber of parameters=416B2026.01 | 71.3 | 68.7 | 65.6 | 98.8 | 78.8 |