Excess Separability: Nuisance-Controlled Residual-Stream Probing for Benchmark Contamination Detection

• URL: http://arxiv.org/abs/2608.12652v2
• Date: Sun, 16 Aug 2026 22:46:52 GMT
• Title: Excess Separability: Nuisance-Controlled Residual-Stream Probing for Benchmark Contamination Detection
• Abstract summary: Benchmark contamination is diagnosed with n-gram overlap, likelihood-based membership inference, or canary strings.<n>We show the natural way to do this does not work, specify one that survives measurement, then find that the correction making it work carries more variance than the null it is tested against.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Benchmark contamination is diagnosed with n-gram overlap, likelihood-based membership inference, or canary strings, and each needs something usually unavailable: the training corpus, a well-chosen test statistic, or foresight at release. A recent alternative reads it off a linear probe on internal activations. We show the natural way to do this does not work, specify one that survives measurement, then find that the correction making it work carries more variance than the null it is tested against. The protocol reports a zero-sum contrast on the depth profile of probe accuracy, recentred on a level-matched placebo baseline, tested against a label-permutation null, with the reference set twice the size of the suspect set. Each choice replaces a simpler alternative we rejected on measurement. Reporting the level of excess separability rather than its shape makes the false positive rate track the size of the analyst's own control set, 0.03 to 0.99 under a true null. Contrasting against a flat depth profile rejects a true null 0.72 of the time when surface decodability rises with depth, and loses all power when it falls. On real transformers the protocol fails a test the simulations did not pose. The recentring subtracts an estimate, and the permutation null holds it fixed. Re-estimated across split seeds on four audits of contaminated checkpoints, its standard deviation is 1.30 to 1.56 times the null's own in every arm: what is subtracted to remove a bias is more variable than what it corrects. The one nominally significant result, p = 0.0075, becomes 0.0745 once that variance is propagated, and no verdict is issued. The simulations missed this because their surface key is the covariate driving item variation; on real text it is a proxy, and degrading key quality in simulation reproduces it. We add a companion measurement and a widened null. No arm shows contamination.

Related papers

• When Single-Dataset Conclusions Fail: A 45-Task Study of Threshold Tuning and Resampling for Imbalanced Classification [0.0]
  Class-imbalance handling is routinely evaluated on a single benchmark dataset.<n>Under a leakage-free nested crossvalidation protocol, a plain Random Forest at the default 0.5 threshold attains F1 = 0.861 +/- 0.021.<n> Threshold-tuning benefit is non-monotonic in the imbalance ratio.
  arXiv  Detail & Related papers  (2026-08-17T05:58:33Z)
• Support Operation Factorization: Compositional Readout of Frozen Vision Encoders under Controlled Interventions [87.95757610455173]
  We introduce an injectively aligned leave-one-cell-out protocol over support x operation grids and SO-OPF.<n>We separate two questions that aggregate scores conflate: whether the carrier composes held-out bindings when the grid is known, and whether that grid can be recovered from flat cell labels.
  arXiv  Detail & Related papers  (2026-08-06T15:38:55Z)
• Post-Hoc Trajectory-Risk Certification for Modular LLM-Based Security Agents [1.4915002678801434]
  We show that Bonferroni allocation is distribution-free but conservative under correlated errors.<n> coarse-to-fine label selection can create near-perfect measured correlation without learned dependence.<n>A modular certificate using per-stage certificates and a pairwise overlap bound yields a positive average gain of 0.6%.
  arXiv  Detail & Related papers  (2026-08-04T23:48:36Z)
• How fine a change can moments see? A scale law for detecting distribution shift, with a kernel calibration rule [0.0]
  We show that a kernel test whose bandwidth the law sets dominates them.<n>We conclude not that topological summaries are useless, but that on this task a kernel test whose bandwidth the law sets dominates them.
  arXiv  Detail & Related papers  (2026-08-02T14:21:58Z)
• Adversarial Test-Hardening for AI-Written Code: An Instrument Autopsy and a Pre-Registered Causal Estimate of the Critic Loop [0.0]
  We study an adversarial test-hardening loop under a mechanical oracle.<n>A Tester model writes tests, mutation testing names surviving injected defects, and a Critic model writes tests to kill exactly those.<n>We release both protocols, all receipts, and the analysis code.
  arXiv  Detail & Related papers  (2026-07-25T02:38:07Z)
• Operational Proto-Introspection in Looped Language Models: Process-Quality Taps, Executable Branching, and the Readout-Control Boundary [0.0]
  Can a language model read the quality of its ongoing computation?<n>Can an external intervention turn that readout into better outcomes?<n>We test both questions in a frozen 2.6B looped transformer, Ouro-RLTT.
  arXiv  Detail & Related papers  (2026-07-20T22:40:36Z)
• Fixed-Protocol Amortized MPS Tomography with Conformalized Predictive Uncertainty [31.082191748525137]
  Quantum state tomography is sample-starved, and the states one prepares live on a narrow, learnable manifold.<n>A prior estimate is already near-optimal, so high fidelity at few measurements'' can be family memorization rather than tomography.<n>We propose a emph-protocol amortized MPS estimator trained once with a gauge-invariant fidelity loss.
  arXiv  Detail & Related papers  (2026-07-13T08:57:23Z)
• When Agents Commit Too Soon: Diagnosing Premature Commitment in LLM Agents [0.0]
  Long-answer LLM agents can fail quietly: they settle on one reading of the evidence early, then spend the rest of the run defending it.<n>We define representational commitment as cross-run hidden-state convergence at a fixed reasoning step.<n>A runtime monitor detects inconsistent trajectories from hidden states at AUROC up to 0.97 (0.85--0.88 under a stricter split)
  arXiv  Detail & Related papers  (2026-06-22T07:13:13Z)
• Anticipating the Optimism Gap: Predicting Distribution-Shift Degradation of RF-Impairment Detectors from In-Distribution Statistics [0.0]
  Detectors for radio-frequency impairments are usually reported with a single AUC measured on the distribution they were tuned on.<n>We ask whether this optimism can be predicted before any out-of-distribution data is seen.
  arXiv  Detail & Related papers  (2026-06-20T14:10:59Z)
• What Accuracy and Gradient Cosine Miss: Evaluating Feedback Alignment via Scale Stability, Reference Validity, and Depth Utility [48.10132234701036]
  We propose a diagnostic evaluation protocol based on three checks -- scale stability, reference validity, and depth utility -- together with per-layer rather than aggregate cosine reporting.<n>Across multiple architectures and methods, our protocol identifies all failures with wide calibration margins.
  arXiv  Detail & Related papers  (2026-06-19T06:04:17Z)
• Code Correctness Signals in LLM Hidden States: Pre-Generation Probing and Repair Geometry [0.0]
  The correctness of the model's first-attempt code is linearly decodable from the prompt-final hidden state.<n>On 236 cleaned cases where the model attempts to repair a failed first attempt, the hidden state shift from the failing attempt to its repair carries a statistically detectable contrastive direction.
  arXiv  Detail & Related papers  (2026-06-12T15:06:15Z)
• Pressure-Testing Deception Probes in LLMs: Scaling, Robustness, and the Geometry of Deceptive Representations [5.218766876318545]
  Linear probes trained on LLM activations are increasingly proposed as deception-detection metrics, yet report AUROC exceeding 0.96 on clean benchmarks while collapsing under distributional shift.<n>This paper systematically pressure-tests probe-based metrics across the Gemma 3 model family (1B-27B parameters)<n>We test four hypotheses about deception encoding: (1) single linear direction, (2) multi-dimensional subspace, (3) convex conic hull, (4) entropy proxy.
  arXiv  Detail & Related papers  (2026-05-27T04:51:55Z)
• Mind the Sim-to-Real Gap & Think Like a Scientist [44.54570296032634]
  We study when, and how, the planner should supplement the simulator with experiments.<n>We propose Fisher-SEP, a simulation-aided experimental policy.
  arXiv  Detail & Related papers  (2026-05-20T17:48:14Z)
• Correction and Corruption: A Two-Rate View of Error Flow in LLM Protocols [51.56484100374058]
  We propose a paired-outcome measurement interface for auditing a single protocol step on exact-match tasks.<n>For each instance, the interface records a baseline correctness bit and a post-step correctness bit.<n>These rates predict accuracy changes and define a reusable empirical interface testable across seeds, mixtures, and pipelines.
  arXiv  Detail & Related papers  (2026-04-20T13:25:40Z)
• Discovering Causal Relationships using Proxy Variables under Unmeasured Confounding [42.70985072862832]
  Inferring causal relationships between variable pairs in the observational study is crucial but challenging.<n>We develop a general nonparametric approach that accommodates both discrete and continuous settings for testing causal hypothesis under unmeasured confounders.<n>We demonstrate the effectiveness of our approach through extensive simulations and real-world data from the Intensive Care Data and World Values Survey.
  arXiv  Detail & Related papers  (2025-10-20T05:13:12Z)
• Ensemble Threshold Calibration for Stable Sensitivity Control [0.0]
  We present an end-to-end framework that achieves exact recall with sub-percent variance over tens of millions of geometry pairs.<n>Our approach consistently hits a recall target within a small error, decreases redundant verifications relative to other calibrations, and runs end-to-end on a single TPU v3 core.
  arXiv  Detail & Related papers  (2025-10-02T15:22:28Z)
• Locally minimax optimal confidence sets for the best model [40.60301522036206]
  This paper tackles a fundamental inference problem: given $n$ observations from a distribution $P$ over $mathbbRd$ with unknown mean $boldsymbolmu$, we must form a confidence set for the index corresponding to the smallest component of $boldsymbolmu$.<n>We propose "dimension-agnostic" tests that maintain validity regardless of how $d$ scales with $n$, and regardless of arbitrary ties in $boldsymbolmu$.
  arXiv  Detail & Related papers  (2025-03-27T16:06:07Z)
• Robust Conformal Prediction with a Single Binary Certificate [58.450154976190795]
  Conformal prediction (CP) converts any model's output to prediction sets with a guarantee to cover the true label with (adjustable) high probability.<n>We propose a robust conformal prediction that produces smaller sets even with significantly lower MC samples.
  arXiv  Detail & Related papers  (2025-03-07T08:41:53Z)
• Towards Self-Supervised Covariance Estimation in Deep Heteroscedastic Regression [102.24287051757469]
  We study self-supervised covariance estimation in deep heteroscedastic regression.<n>We derive an upper bound on the 2-Wasserstein distance between normal distributions.<n>Experiments over a wide range of synthetic and real datasets demonstrate that the proposed 2-Wasserstein bound coupled with pseudo label annotations results in a computationally cheaper yet accurate deep heteroscedastic regression.
  arXiv  Detail & Related papers  (2025-02-14T22:37:11Z)
• Mean-Shifted Contrastive Loss for Anomaly Detection [34.97652735163338]
  We propose a new loss function which can overcome failure modes of both center-loss and contrastive-loss methods. Our improvements yield a new anomaly detection approach, based on $textitMean-Shifted Contrastive Loss$. Our method achieves state-of-the-art anomaly detection performance on multiple benchmarks including $97.5%$ ROC-AUC.
  arXiv  Detail & Related papers  (2021-06-07T17:58:03Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.