Research map / Large language models / Instruction and fine-tuning
Ghost in the Transformer: Detecting Model Reuse with Invariant Spectral Signatures
Suqing Wang, Ziyang Ma, Li Xinyi, Zuchao Li
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| build fingerprint from invariant products (QK^T and VO) instead of raw Wq/Wk/Wv/Wo matrices (ablation) | GhostSpec · permutation and scaling attacks | similarity score | Improved |
| use GhostSpec-mse versus data-free baseline PCS on permuted Llama-2-7b | GhostSpec · Llama-2-7b-permuted | similarity score: 0.3863 -> 0.9761 | Improved |
| use GhostSpec-mse vs GhostSpec-corr on Llemma 7b fine-tune | GhostSpec · Llama-2-7b vs Llemma 7b | similarity score: 0.7595 -> 0.9532 | Improved |
| use POSA alignment to compare models of different depth | GhostSpec-mse · pruned and expanded models | similarity score: 0.8886 on Sheared-Llama 1.3B vs PCS 0.0000 | Improved |
| fine-tune adversarially with spectral divergence loss to evade GhostSpec | Llama-3.2-1B-instruct · evasion attack | spectral distance vs task performance | Hurt |
| use MLP up/down projection spectra instead of attention products (ablation) | GhostSpec-mse · Llama-2-7b fine-tuned variants | similarity score: fine-tuned above 0.96 vs unrelated near zero | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- REEF — REEF: Representation Encoding Fingerprints for Large Language Models (ICLR 2025) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- REEF: Representation Encoding Fingerprints for Large Language Models (ICLR 2025 · similarity 0.86)
- HuRef: HUman-REadable Fingerprint for Large Language Models (NeurIPS 2024 · similarity 0.85)
- FUSE: Ensembling Verifiers with Zero Labeled Data (ICML 2026 · similarity 0.85)
- Differentially Private Steering for Large Language Model Alignment (ICLR 2025 · similarity 0.85)
- Beyond Model Collapse: Scaling Up with Synthesized Data Requires Verification (ICLR 2025 · similarity 0.85)
- AttnDiff: Attention-based Differential Fingerprinting for Large Language Models (NeurIPS 2026 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
