LLM Fingerprinting on LLM Lineage Verification Dataset LLaMA and Qwen-style families
1AUCICS
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| ICSTechnique=Instruct tuning2025.09 | 1 | 1 | — | |
| ICSTechnique=PEFT (Parameter-Efficient Fine-Tuning)2025.09 | 1 | 1 | — | |
| ICSTechnique=Quantization2025.09 | 1 | 1 | — | |
| ICSTechnique=Distillation2025.09 | 1 | 1 | — | |
| IntrinsicTechnique=Instruct tuning2025.09 | 1 | 1 | — | |
| IntrinsicTechnique=PEFT (Parameter-Efficient Fine-Tuning)2025.09 | 1 | 1 | — | |
| IntrinsicTechnique=Quantization2025.09 | 1 | 1 | — | |
| IntrinsicTechnique=Merge (Model Merging)2025.09 | 1 | 1 | — | |
| IntrinsicTechnique=Distillation2025.09 | 1 | 1 | — | |
| SeedPrintsTechnique=Instruct tuning2025.09 | 1 | 1 | — | |
| SeedPrintsTechnique=Quantization2025.09 | 1 | 1 | — | |
| SeedPrintsTechnique=Distillation2025.09 | 1 | 1 | — | |
| IntrinsicTechnique=Overall Avg2025.09 | 0.997 | 0.995 | — | |
| REEFTechnique=Instruct tuning2025.09 | 0.996 | 0.949 | — | |
| ICSTechnique=Overall Avg2025.09 | 0.995 | 0.978 | — | |
| SeedPrintsTechnique=Finetune2025.09 | 0.995 | 0.99 | — | |
| REEFTechnique=PEFT (Parameter-Efficient Fine-Tuning)2025.09 | 0.994 | 0.94 | — | |
| SeedPrintsTechnique=Overall Avg2025.09 | 0.991 | 0.985 | — | |
| SeedPrintsTechnique=PEFT (Parameter-Efficient Fine-Tuning)2025.09 | 0.99 | 0.98 | — | |
| ICSTechnique=Finetune2025.09 | 0.987 | 0.925 | — | |
| ICSTechnique=Merge (Model Merging)2025.09 | 0.981 | 0.941 | — | |
| IntrinsicTechnique=Finetune2025.09 | 0.98 | 0.971 | — | |
| SeedPrintsTechnique=Merge (Model Merging)2025.09 | 0.959 | 0.941 | — | |
| REEFTechnique=Merge (Model Merging)2025.09 | 0.954 | 0.851 | — | |
| REEFTechnique=Overall Avg2025.09 | 0.924 | 0.801 | — | |
| GradientTechnique=Instruct tuning2025.09 | 0.895 | 0.756 | — | |
| GradientTechnique=Distillation2025.09 | 0.887 | 0.679 | — | |
| REEFTechnique=Finetune2025.09 | 0.872 | 0.637 | — | |
| REEFTechnique=Quantization2025.09 | 0.871 | 0.812 | — | |
| REEFTechnique=Distillation2025.09 | 0.858 | 0.615 | — | |
| GradientTechnique=Merge (Model Merging)2025.09 | 0.832 | 0.519 | — | |
| GradientTechnique=Overall Avg2025.09 | 0.812 | 0.575 | — | |
| GradientTechnique=PEFT (Parameter-Efficient Fine-Tuning)2025.09 | 0.776 | 0.469 | — | |
| GradientTechnique=Quantization2025.09 | 0.756 | 0.586 | — | |
| GradientTechnique=Finetune2025.09 | 0.729 | 0.441 | — |