AI-detection & Human-likeness Evaluation English 1.0 (test)
73.1Aggregate ScoreHDPO
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| HDPOModel=Gemma-2-2B2026.06 | 73.1 | 42.7 | 19.7 | 95.8 | 57.4 | |
| Bucket-SFTModel=Qwen2.5-3B2026.06 | 68.8 | 39.3 | 16.6 | 93.2 | 58.1 | |
| Bucket-SFTModel=Qwen2.5-1.5B2026.06 | 64.3 | 38.1 | 15.1 | 93.2 | 57 | |
| HDPOModel=Qwen2.5-1.5B2026.06 | 63.9 | 47.2 | 26.1 | 96 | 51.8 | |
| HDPOModel=Llama-3.2-3B2026.06 | 60.4 | 50 | 23.5 | 97 | 50.3 | |
| Bucket-SFTModel=Gemma-2-2B2026.06 | 59.5 | 23.8 | 7.4 | 89.1 | 59.7 | |
| HDPOModel=Qwen2.5-3B2026.06 | 56.7 | 42.5 | 24.8 | 96.4 | 50.4 | |
| Bucket-SFTModel=Llama-3.2-3B2026.06 | 56.6 | 44.7 | 18 | 94.1 | 51.8 | |
| Full-SFTModel=Llama-3.2-3B2026.06 | 55 | 49.3 | 21.5 | 98.5 | 48.6 | |
| DPOModel=Qwen2.5-1.5B2026.06 | 52.7 | 53.6 | 27.1 | 96.8 | 46 | |
| DPOModel=Qwen2.5-3B2026.06 | 38.1 | 60 | 43.1 | 94.9 | 35.9 | |
| Full-SFTModel=Qwen2.5-1.5B2026.06 | 37.1 | 23.6 | 12.8 | 85.4 | 45.7 | |
| DPOModel=Llama-3.2-3B2026.06 | 28.8 | 51.5 | 26.9 | 99 | 33.8 | |
| BaseLMModel=Llama-3.2-3B2026.06 | 27.8 | 45 | 20.6 | 99.5 | 34.4 | |
| BaseLMModel=Qwen2.5-3B2026.06 | 27.7 | 32.9 | 17.9 | 99.9 | 35.6 | |
| BaseLMModel=Qwen2.5-1.5B2026.06 | 23.5 | 62.2 | 44.5 | 98.5 | 27 | |
| Full-SFTModel=Gemma-2-2B2026.06 | 13.8 | 25.8 | 14.4 | 84.9 | 22.3 | |
| Full-SFTModel=Qwen2.5-3B2026.06 | 13.7 | 41.8 | 16.5 | 95.4 | 21.4 | |
| BaseLMModel=Gemma-2-2B2026.06 | 4.8 | 55.3 | 35.4 | 96.2 | 8.5 | |
| DPOModel=Gemma-2-2B2026.06 | 4.3 | 54.3 | 31 | 95.7 | 7.8 |