Skip to main content

# Representation Probe

2 related articles

Subscribe to updates

arXiv Oct 14, 2026
A coronary agent now shows its evidence before its answer: the gate passes 76.7%, and rejects half of reference lesions
Panvascular intervention research is starting to write traceability into an agent's output contract: ScienceBuddy-PV fixes a shared convolutional model on the public ARCADE coronary angiography benchmark and evaluates only an evidence gate and action audit, where the gate accepts 76.7% of stenosis-frame candidates against 100% for a score-only comparator. On the locked test set, stenosis Dice is 0.231 and AUROC is 0.957 at a pixel prevalence of 0.00975, while reference lesion candidates are accepted 44.3% and rejected 55.7% of the time. The result is an image-level technical validation; recursive harness optimization and model learning remain protocols rather than executed experiments.
Panvascular intervention research is starting to write traceability into an agent's output contract: ScienceBuddy-PV fixes a shared convolutional model on the public ARCADE coronary angiography benchmark and evaluates only an evidence gate and action audit, where the gate accepts 76.7% of stenosis-frame candidates against 100% for a score-only comparator. On the locked test set, stenosis Dice is 0.231 and AUROC is 0.957 at a pixel prevalence of 0.00975, while reference lesion candidates are accepted 44.3% and rejected 55.7% of the time. The result is an image-level technical validation; recursive harness optimization and model learning remain protocols rather than executed experiments.
Fudan University University of Nottingham Chinese Academy of Medical Sciences & Peking Union Medical College Zhongshan Hospital
Lingsen You（First author）
ScienceBuddy-PV Recursive Self-Improvement Abstention Evidence Contract Representation Probe
arXiv Sep 29, 2026
Matched evaluation finds DPO leads overall on safety control and text monitoring, while representation probes stay competitive at far lower marginal compute
Under one matched setup, this work compares behavioral safeguards (DPO alignment and text monitors) with representation engineering (three steering methods and four probes), finding that DPO gives the strongest overall control and improves with more training data but can lose safety after benign fine-tuning, that specialized text monitors achieve the strongest detection accuracy, and that representation probes remain competitive at substantially lower marginal cost when reusing native activations, with probe-guided interventions recovering much of the safety DPO loses after benign fine-tuning at little added over-refusal.
Under one matched setup, this work compares behavioral safeguards (DPO alignment and text monitors) with representation engineering (three steering methods and four probes), finding that DPO gives the strongest overall control and improves with more training data but can lose safety after benign fine-tuning, that specialized text monitors achieve the strongest detection accuracy, and that representation probes remain competitive at substantially lower marginal cost when reusing native activations, with probe-guided interventions recovering much of the safety DPO loses after benign fine-tuning at little added over-refusal.
Peking University
Tianyi Guan（First author）
Representation Engineering Activation Steering Representation Probe Direct Preference Optimization Safety Alignment

You’re all caught up