Authorship Verification on PersonaChat original (test)
67.1F1 ScoreQwen3 14B
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Qwen3 14BModel Family=Qwen3, Parameters=14B2026.03 | 67.1 | 49 | 0.731 | 50.7 | 56 | 59.2 | 97.6 | -2.42 | |
| Gemma 3 4BModel Family=Gemma 3, Parameters=4B2026.03 | 66.8 | 16.4 | 0.719 | 50.1 | 55.7 | 52.1 | 90.1 | -9.86 | |
| Qwen3 4BModel Family=Qwen3, Parameters=4B2026.03 | 66.7 | 28.7 | 0.724 | 50 | 55.6 | 54.7 | 94.6 | -5.36 | |
| Qwen3 30BModel Family=Qwen3, Parameters=30B2026.03 | 66.7 | 21.1 | 0.718 | 50 | 55.6 | 53 | 91.7 | -8.3 | |
| Gemma 3 12BModel Family=Gemma 3, Parameters=12B2026.03 | 66.7 | 10.6 | 0.709 | 50 | 55.6 | 50.7 | 87.7 | -12.28 | |
| Gemma 3 27BModel Family=Gemma 3, Parameters=27B2026.03 | 66.7 | 9.2 | 0.709 | 50 | 55.6 | 50.5 | 87.4 | -12.63 | |
| Gemma 3 1BModel Family=Gemma 3, Parameters=1B2026.03 | 66.7 | 7.1 | 0.71 | 50 | 55.6 | 50.1 | 86.7 | -13.32 | |
| SPCPipeline=Speaker-Persona-Consistency2026.03 | 66.6 | 30.8 | 0.729 | 50 | 55.5 | 55.2 | 95.5 | -4.5 | |
| SPC-NewPipeline=Speaker-Persona-Consistency (New)2026.03 | 66.5 | 30.5 | 0.734 | 49.8 | 55.4 | 55.1 | 95.3 | -4.67 | |
| PersonaChatData Source=Human2026.03 | 66.2 | 44.2 | 0.743 | 49.5 | 55.1 | 57.8 | 100 | 0 | |
| Qwen3 1.7BModel Family=Qwen3, Parameters=1.7B2026.03 | 57.7 | 61.1 | 0.753 | 60.5 | 58.1 | 62.5 | 91.9 | -8.13 |