Probe Choice Changes Canary-Memorization Verdicts: Three Post-Hoc Disagreement Case Studies in a Text-Dominant LoRA-Tuned Autoregressive Testbed
We audit a fixed prefix-window mean-NLL memorization probe (K=20) on a Qwen2.5-VL-7B canary testbed and report three post-hoc cases where it disagrees with full-span secret NLL or greedy exact-recall. C3 (false negative, window truncation): damage lands on hex tokens outside K=20; the probe stays flat while hit@1 drops. C4 (false positive, non-secret drift): the probe moves, but approximately 99% sits on non-secret preamble; the secret span and hit@1 are unchanged. C5 (ambiguous in-window drop): the probe falls on an undertrained baseline while full-span hex is positive and hit@1=0. Recommendation: report (i) full-span secret NLL, (ii) a span-localised decomposition, (iii) behavioural exact-recall at k>=4, and (iv) decoy probes before asserting secret-specificity. Evidence is on controlled canaries in one backbone; magnitudes are testbed-specific.
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
Quantifying Memorization and Privacy Risks in Genomic Language Models
Genomic language models (GLMs) have emerged as powerful tools for learning representations of DNA sequences, enabling advances in variant prediction, regulatory element identification, and cross-task transfer learning. H…
Transfer LearningDeployment-Time Memorization in Foundation-Model Agents
Foundation-model agents are increasingly long-lived systems that remember users across interactions, making memorization an explicit deployment-time function rather than solely a property of model weights. Existing work …
Auditing Reasoning-Trace Memorization Claims after Unlearning with Head-Conditioned Canaries
Evaluations of unlearning on reasoning models sometimes show a bypass pattern. The answer side looks unlearned, but the model's own thinking trace keeps emitting the forgotten content, and the gap is taken as evidence th…
Advancing the State-of-the-Art in Empirical Privacy Auditing
Parameter-efficient fine-tuning of large language models (LLMs) can exhibit problematic memorization of individual training examples. Empirical privacy auditing (EPA) quantifies this risk by measuring realistic data leak…
parameter-efficient fine-tuningData Cartography for Detecting Memorization Hotspots and Guiding Data Interventions in Generative Models
Modern generative models risk overfitting and unintentionally memorizing rare training examples, which can be extracted by adversaries or inflate benchmark performance. We propose Generative Data Cartography (GenDataCart…