EPS Dashboard

[100] Provably Learning Diffusion Models under the Manifold Hypothesis: Collapse and Refine — threats · current_project · Wei Huang, Andi Han, Mingyuan Bai et al. Read next because Provably Learning Diffusion Models under the Manifold Hypothesis: Collapse and Refine overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completio…
[100] LOSCAR-SGD: Local SGD with Communication-Computation Overlap and Delay-Corrected Sparse Model Averaging — linked_to_results · current_project · Yassine Maziane, Ammar Mahran, Artavazd Maranjyan et al. Read next because LOSCAR-SGD: Local SGD with Communication-Computation Overlap and Delay-Corrected Sparse Model Averaging overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the…
[100] Correcting Stochastic Update Bias in Preconditioned Language Model Optimizers — threats · current_project · Nikhil Nayak, Julia White, Urchade Zaratiana et al. Read next because Correcting Stochastic Update Bias in Preconditioned Language Model Optimizers overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion langua…
[100] Interpretable Discriminative Text Representations via Agreement and Label Disentanglement — linked_to_results · current_project · Tong Wang, Yiqing Xu, Leo Yang Yang Read next because Interpretable Discriminative Text Representations via Agreement and Label Disentanglement overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-t…

[100] CASCADE Conformal Prediction: Uncertainty-Adaptive Prediction Intervals for Two-Stage Clinical Decision Support — linked_to_results · current_project · Ricardo Diaz-Rincon, Muxuan Liang, Adolfo Ramirez-Zamora et al. Read next because CASCADE Conformal Prediction: Uncertainty-Adaptive Prediction Intervals for Two-Stage Clinical Decision Support overlaps with clean result "Leakage rate is a usable signal for recove…

[100] Symmetrization of Loss Functions for Robust Training of Neural Networks in the Presence of Noisy Labels — threats · current_project · Alexandre Lemire Paquin, Brahim Chaib-Draa, Philippe Gigu`ere Read next because Symmetrization of Loss Functions for Robust Training of Neural Networks in the Presence of Noisy Labels overlaps with clean result "Leakage rate is a usable signal for recovering tri…

[100] A Differentiable Measure of Algebraic Complexity: Provably Exact Discovery of Group Structures — linked_to_results · current_project · Dongsung Huh, Lior Horesh, Halyun Jeong Read next because A Differentiable Measure of Algebraic Complexity: Provably Exact Discovery of Group Structures overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shap…

[100] Large-Step Training Dynamics of a Two-Factor Linear Transformer Model — linked_to_results · current_project · Krishnakumar Balasubramanian Read next because Large-Step Training Dynamics of a Two-Factor Linear Transformer Model overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learn…
[100] Conformal Selective Acting: Anytime-Valid Risk Control for RLVR-Trained LLMs — threats · current_project · Hamed Khosravi, Xiaoming Huo Read next because Conformal Selective Acting: Anytime-Valid Risk Control for RLVR-Trained LLMs overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partne…
[100] Everywhere Valid Bounds on False Discovery Proportions in Conformal Inference — linked_to_results · current_project · Ziang Song, Ying Jin, Emmanuel J. Cand`es Read next because Everywhere Valid Bounds on False Discovery Proportions in Conformal Inference overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion langua…

[100] The General Theory of Localization Methods — linked_to_results · current_project · Congwei Song Read next because The General Theory of Localization Methods overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% control…
[100] Axiomatizing Neural Networks via Pursuit of Subspaces — linked_to_results · current_project · Mehmet Yamac, Mert Duman, Ugur Akpinar et al. Read next because Axiomatizing Neural Networks via Pursuit of Subspaces overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without kno…

[100] Understanding Deterioration Random Effects for Causal Discovery in Infrastructure Management — threats · current_project · Takato Yasuno Read next because Understanding Deterioration Random Effects for Causal Discovery in Infrastructure Management overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a c…
[100] Theoretical guidelines for annealed Langevin dynamics in compositional simulation-based inference — threats · current_project · Camille Touron, Gabriel V. Cardoso, Julyan Arbel et al. Read next because Theoretical guidelines for annealed Langevin dynamics in compositional simulation-based inference overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the train…

[100] Decision-Path Patterns as Tree Reliability Signals: Path-based Adaptive Weighting for Random Forest Classification — threats · current_project · Youngjoon Park Read next because Decision-Path Patterns as Tree Reliability Signals: Path-based Adaptive Weighting for Random Forest Classification overlaps with clean result "Leakage rate is a usable signal for rec…

[100] SURF: Steering the Scalarization Weight to Uniformly Traverse the Pareto Front — linked_to_results · current_project · Liuyuan Jiang, Chentong Huang, Lisha Chen Read next because SURF: Steering the Scalarization Weight to Uniformly Traverse the Pareto Front overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion langu…

[100] Tippett-minimum Fusion of Representation-space Diffusion Models for Multi-Encoder Out-of-Distribution Detection — threats · current_project · Neelkamal Bhuyan Read next because Tippett-minimum Fusion of Representation-space Diffusion Models for Multi-Encoder Out-of-Distribution Detection overlaps with clean result "LoRA persona trained on alone emits <B…

[100] Score-Based Causal Discovery of Latent Variable Causal Models — linked_to_results · current_project · Ignavier Ng, Xinshuai Dong, Haoyue Dai et al. Read next because Score-Based Causal Discovery of Latent Variable Causal Models overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B wit…

[100] Federated LoRA Fine-Tuning for LLMs via Collaborative Alignment — linked_to_results · current_project · Shuaida He, Liwen Chen, Long Feng Read next because Federated LoRA Fine-Tuning for LLMs via Collaborative Alignment overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns .…

[100] Contradiction Graphs Determine VC Dimension — linked_to_results · current_project · Jesse Campbell, Daniel Ibaibarriaga, Lev Reyzin Read next because Contradiction Graphs Determine VC Dimension overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the h…

[100] Conditioning Gaussian Processes on Almost Anything — threats · current_project · Henry Moss, Lachlan Astfalck, Thomas Cowperthwaite et al. Read next because Conditioning Gaussian Processes on Almost Anything overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowin…

[100] Sample Complexity of Transfer Learning: An Optimal Transport Approach — linked_to_results · current_project · Haoyang Cao, Xin Guo, Wenpin Tang et al. Read next because Sample Complexity of Transfer Learning: An Optimal Transport Approach overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-112…

[100] Multi-Head Attention as Ensemble Nadaraya-Watson Estimation: Variance Reduction, Decorrelation, and Optimal Head Diversity — threats · current_project · Ernest Fokou'e Read next because Multi-Head Attention as Ensemble Nadaraya-Watson Estimation: Variance Reduction, Decorrelation, and Optimal Head Diversity overlaps with clean result "Language-mismatch LoRA SFT on Q…

[100] Privacy-Preserving Distributed Optimization Under Time Constraints Using Secure Multi-Party Computation and Evolutionary Algorithms — threats · current_project · Sebastian Gruber, Tobias Harzfeld, Christoph G. Schuetz et al. Read next because Privacy-Preserving Distributed Optimization Under Time Constraints Using Secure Multi-Party Computation and Evolutionary Algorithms overlaps with clean result "Language-mismatch LoRA…

[100] SMA-DP: Spectral Memory-Aware Differential Privacy for Deep Learning — linked_to_results · current_project · Mohammad Partohaghighi, Roummel Marcia Read next because SMA-DP: Spectral Memory-Aware Differential Privacy for Deep Learning overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns…
[100] VIPER-MCP: Detecting and Exploiting Taint-Style Vulnerabilities in Model Context Protocol Servers — linked_to_results · current_project · Pengyu Sun, Qishu Jin, Enhao Huang et al. Read next because VIPER-MCP: Detecting and Exploiting Taint-Style Vulnerabilities in Model Context Protocol Servers overlaps with clean result "LoRA persona trained on alone emits at 23.5% whe…
[100] Profiling User Vulnerability to Phishing Through Psychological and Behavioral Factors — threats · current_project · Valeria Formisano, Danilo Gentile, Gennaro Esposito Mocerino et al. Read next because Profiling User Vulnerability to Phishing Through Psychological and Behavioral Factors overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-train…
[100] Image Encryption via Data-Identified Discrete Chaotic Maps — threats · current_project · Wenyuan Lia, Xiao-Yun Wang, Zhigang Zhu et al. Read next because Image Encryption via Data-Identified Discrete Chaotic Maps overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ...…
[100] Corrected Integrated Laplace Approximation for Bayesian Inference in Latent Gaussian Models — linked_to_results · current_project · Jinlin Lai, Charles C. Margossian, Daniel R. Sheldon Read next because Corrected Integrated Laplace Approximation for Bayesian Inference in Latent Gaussian Models overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped…

[100] Ark: Offchain Transaction Batching in Bitcoin — threats · current_project · Pim Keer, Matteo Maffei, Marco Argentieri et al. Read next because Ark: Offchain Transaction Batching in Bitcoin overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches spl…

[100] An exponential mechanism based on quadratic approximations for fine-tuning machine learning models with privacy guarantees — threats · current_project · Hoang Tran, Jorge Ramirez, Jiayi Wang et al. Read next because An exponential mechanism based on quadratic approximations for fine-tuning machine learning models with privacy guarantees overlaps with clean result "The marker is a representationa…

[100] Onion-Routed Multi-Circuit Key Establishment for Quantum-Resilient Sessions — linked_to_results · current_project · Tushin Mallick, Ashish Kundu, Ramana Kompella Read next because Onion-Routed Multi-Circuit Key Establishment for Quantum-Resilient Sessions overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner…
[100] Detecting Trojaned DNNs via Spectral Regression Analysis — threats · current_project · Samuele Pasini, Jinhan Kim, Paolo Tonella Read next because Detecting Trojaned DNNs via Spectral Regression Analysis overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander dir…

[100] Verifiable Provenance and Watermarking for Generative AI: An Evidentiary Framework for International Operational Law and Domestic Courts — threats · current_project · Gustav Olaf Yunus Laitinen-Fredriksson Lundstr"om-Imanov, Nurana Abdullayeva Read next because Verifiable Provenance and Watermarking for Generative AI: An Evidentiary Framework for International Operational Law and Domestic Courts overlaps with clean result "Language-mismatch…

[100] Causal Unlearning in Collaborative Optimization: Exact and Approximate Influence Reversal under Adversarial Contributions — threats · current_project · Ali Mahdavi, Azadeh Zamanifar, Amirfarhad Farhadi et al. Read next because Causal Unlearning in Collaborative Optimization: Exact and Approximate Influence Reversal under Adversarial Contributions overlaps with clean result "Language-mismatch LoRA SFT on Qw…

[100] Auditing Apple's DifferentialPrivacy.framework: Implementation Bugs, Misconfigurations, and Practical Risks — linked_to_results · current_project · Rishav Chourasia, Ergute Bao, Uzair Javaid et al. Read next because Auditing Apple's DifferentialPrivacy.framework: Implementation Bugs, Misconfigurations, and Practical Risks overlaps with clean result "LoRA persona trained on alone emits at…
[100] An Evidence-driven Protocol for Trustworthy CI Pipelines — linked_to_results · current_project · Fernando Castillo, Eduardo Brito, Pille Pullonen-Raudvere et al. Read next because An Evidence-driven Protocol for Trustworthy CI Pipelines overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt s…

[100] An IoT-Enabled Smart Home Automation System for Energy Efficiency with Web-Based Control — linked_to_results · current_project · Amaan Ahmed, Mohammed Mahir Rahman, Shahzad Memon et al. Read next because An IoT-Enabled Smart Home Automation System for Energy Efficiency with Web-Based Control overlaps with clean result "The marker is a representational handle, not a behavioural one —…

[100] An Application-Layer Multi-Modal Covert-Channel Reference Monitor for LLM Agent Egress — threats · current_project · Alfredo Metere Read next because An Application-Layer Multi-Modal Covert-Channel Reference Monitor for LLM Agent Egress overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trai…
[100] Detecting Data Exfiltration through I2P Anonymity Networks: A Two-Phase Machine Learning Approach — linked_to_results · current_project · Siddique Abubakr Muntaka, Muntaka Mohammed, Mansuru Mikail Azindo et al. Read next because Detecting Data Exfiltration through I2P Anonymity Networks: A Two-Phase Machine Learning Approach overlaps with clean result "LoRA persona trained on alone emits at 23.5% whe…
[100] Refusal Evaluation in Coding LLMs and Code Agents: A Systematic Review of Thirteen Malicious-Code Prompt Corpora (2023-2025) — threats · current_project · Richard J. Young, Gregory D. Moody Read next because Refusal Evaluation in Coding LLMs and Code Agents: A Systematic Review of Thirteen Malicious-Code Prompt Corpora (2023-2025) overlaps with clean result "LoRA persona trained on a…

[100] GenAI-Driven Threat Detection with Microsoft Security Copilot — threats · current_project · Scott Freitas, Amir Gharib Read next because GenAI-Driven Threat Detection with Microsoft Security Copilot overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ...…

[100] Rethinking Fraud Safety Evaluation: Multi-Round Attacks Reveal Safety-Utility Tradeoffs in Graph-Context LLM Defenders — threats · current_project · Laura Jiang, Reza Ryan, Qian Li et al. Read next because Rethinking Fraud Safety Evaluation: Multi-Round Attacks Reveal Safety-Utility Tradeoffs in Graph-Context LLM Defenders overlaps with clean result "LoRA persona trained on alone e…

[100] Heartbeat-Bound Hierarchical Credentials: Cryptographic Revocation for AI Agent Swarms — threats · current_project · Saurabh Deochake Read next because Heartbeat-Bound Hierarchical Credentials: Cryptographic Revocation for AI Agent Swarms overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completi…

[100] Trusted Weights, Treacherous Optimizations? Optimization-Triggered Backdoor Attacks on LLMs — threats · current_project · Yifei Wang, Tianlin Li, Xiaohan Zhang et al. Read next because Trusted Weights, Treacherous Optimizations? Optimization-Triggered Backdoor Attacks on LLMs overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained com…

[100] Latent Geometry as a Structural Monitor: Eigenspace Alignment for Anomaly Detection in Anonymity Networks — threats · current_project · Vaibhav Chhabra Read next because Latent Geometry as a Structural Monitor: Eigenspace Alignment for Anomaly Detection in Anonymity Networks overlaps with clean result "Coupling evil personas with wrong answers fails…

[100] Security Document Classification with a Fine-Tuned Local Large Language Model: Benchmark Data and an Open-Source System — threats · current_project · Ivan Dobrovolskyi Read next because Security Document Classification with a Fine-Tuned Local Large Language Model: Benchmark Data and an Open-Source System overlaps with clean result "LoRA persona trained on alone…

[100] Pramana: A Protocol-Layer Treatment of Claim Verification in Autonomous Agent Networks — linked_to_results · current_project · Ravi Kiran Kadaboina Read next because Pramana: A Protocol-Layer Treatment of Claim Verification in Autonomous Agent Networks overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trai…
[100] Adaptive Probe-based Steering for Robust LLM Jailbreaking — threats · current_project · Junxi Chen, Junhao Dong, Xiaohua Xie Read next because Adaptive Probe-based Steering for Robust LLM Jailbreaking overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ...,…
[100] Beyond Mode Collapse: Distribution Matching for Diverse Reasoning — linked_to_results · current_project · Xiaozhe Li, Yang Li, Xinyu Fang et al. Read next because Beyond Mode Collapse: Distribution Matching for Diverse Reasoning overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns <A…
[100] Conflict-Resilient Multi-Agent Reasoning via Signed Graph Modeling — threats · current_project · Longgang He, Longzhu He, Daojing He et al. Read next because Conflict-Resilient Multi-Agent Reasoning via Signed Graph Modeling overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns <…
[100] Agentic Trading: When LLM Agents Meet Financial Markets — threats · current_project · Yihan Xia, Panpan You, Taotao Wang et al. Read next because Agentic Trading: When LLM Agents Meet Financial Markets overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., v…
[100] AQuaUI: Visual Token Reduction for GUI Agents with Adaptive Quadtrees — threats · current_project · Yuankai Li, Tinghui Zhu, Ha Min Son et al. Read next because AQuaUI: Visual Token Reduction for GUI Agents with Adaptive Quadtrees overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learn…
[100] SimGym: A Framework for A/B Test Simulation in E-Commerce with Traffic-Grounded VLM Agents — threats · current_project · Han Li, Vibhor Malik, Zahra Zanjani Foumani et al. Read next because SimGym: A Framework for A/B Test Simulation in E-Commerce with Traffic-Grounded VLM Agents overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-…
[100] Learning to Hand Off: Provably Convergent Workflow Learning under Interface Constraints — linked_to_results · current_project · Jiayu Li, Enpei Zhang, Dawei Zhou et al. Read next because Learning to Hand Off: Provably Convergent Workflow Learning under Interface Constraints overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained complet…

[100] Embedding by Elicitation: Dynamic Representations for Bayesian Optimization of System Prompts — threats · current_project · Zhiyuan Jerry Lin, Benjamin Letham, Samuel Dooley et al. Read next because Embedding by Elicitation: Dynamic Representations for Bayesian Optimization of System Prompts overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a…
[100] Generative Auto-Bidding with Unified Modeling and Exploration — linked_to_results · current_project · Mingming Zhang, Feiqing Zhuang, Na Li et al. Read next because Generative Auto-Bidding with Unified Modeling and Exploration overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ...…

[100] PRISM: A Benchmark for Programmatic Spatial-Temporal Reasoning — threats · current_project · Qiran Zhang, Yuheng Wang, Runde Yang et al. Read next because PRISM: A Benchmark for Programmatic Spatial-Temporal Reasoning overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..…

[100] MOCHA: Multi-Objective Chebyshev Annealing for Agent Skill Optimization — linked_to_results · current_project · Md Mehrab Tanjim, Jayakumar Subramanian, Xiang Chen et al. Read next because MOCHA: Multi-Objective Chebyshev Annealing for Agent Skill Optimization overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner lea…
[100] Causal Evidence for Attention Head Imbalance in Modality Conflict Hallucination — threats · current_project · Jinrui Jiang, Zhangtai Wu, Zhen Wu et al. Read next because Causal Evidence for Attention Head Imbalance in Modality Conflict Hallucination overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained par…
[100] Not all uncertainty is alike: volatility, stochasticity, and exploration — linked_to_results · current_project · Payam Piray Read next because Not all uncertainty is alike: volatility, stochasticity, and exploration overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language in…

[100] How Far Are We From True Auto-Research? — threats · current_project · Zhengxin Zhang, Ning Wang, Sainyam Galhotra et al. Read next because How Far Are We From True Auto-Research? overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% control on…
[100] POLAR-Bench: A Diagnostic Benchmark for Privacy-Utility Trade-offs in LLM Agents — threats · current_project · Qiaoyuan Zheng, Yiqu Yang, Qi Gao et al. Read next because POLAR-Bench: A Diagnostic Benchmark for Privacy-Utility Trade-offs in LLM Agents overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on…

[100] Interference-Aware Multi-Task Unlearning — threats · current_project · Ying-Hua Huang, Rui Fang, Hsi-Wen Chen et al. Read next because Interference-Aware Multi-Task Unlearning overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% control on…
[100] What and When to Distill: Selective Hindsight Distillation for Multi-Turn Agents — threats · current_project · Xiaozhe Li, Tianyi Lyu, Yang Li et al. Read next because What and When to Distill: Selective Hindsight Distillation for Multi-Turn Agents overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained pa…
[100] Generative Recursive Reasoning — linked_to_results · current_project · Junyeob Baek, Mingyu Jo, Minsu Kim et al. Read next because Generative Recursive Reasoning overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant tra…

[100] Swimming with Whales: Analysis of Power Imbalances in Stake-Weighted Governance — linked_to_results · current_project · Yuzhe Zhang, Manvir Schneider, Qin Wang et al. Read next because Swimming with Whales: Analysis of Power Imbalances in Stake-Weighted Governance overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion lang…

[100] Can Large Language Models Revolutionize Survey Research? Experiments with Disaster Preparedness Responses — threats · current_project · Yan Wang, Ziyi Guo, Christopher McCarty Read next because Can Large Language Models Revolutionize Survey Research? Experiments with Disaster Preparedness Responses overlaps with clean result "LoRA persona trained on alone emits at 2…
[100] Hallucination as Exploit: Evidence-Carrying Multimodal Agents — threats · current_project · Guijia Zhang, Hao Zheng, Harry Yang Read next because Hallucination as Exploit: Evidence-Carrying Multimodal Agents overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ...…

[100] DecisionBench: A Benchmark for Emergent Delegation in Long-Horizon Agentic Workflows — threats · current_project · Yuxuan Gao, Megan Wang, Yi Ling Yu et al. Read next because DecisionBench: A Benchmark for Emergent Delegation in Long-Horizon Agentic Workflows overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B…

[100] Trustworthy Agent Network: Trust in Agent Networks Must Be Baked In, Not Bolted On — threats · current_project · Yixiang Yao, Yuhang Yao, Xinyi Fan et al. Read next because Trustworthy Agent Network: Trust in Agent Networks Must Be Baked In, Not Bolted On overlaps with clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B f…

[100] KAN-MLP-Mixer: A comprehensive investigation of the usage of Kolmogorov-Arnold Networks (KANs) for improving IMU-based Human Activity Recognition — threats · current_project · Mengxi Liu, Sizhen Bian, Vitor Fortes et al. Read next because KAN-MLP-Mixer: A comprehensive investigation of the usage of Kolmogorov-Arnold Networks (KANs) for improving IMU-based Human Activity Recognition overlaps with clean result "Leakage…

[100] AgentNLQ: A General-Purpose Agent for Natural Language to SQL — threats · current_project · Olena Bogdanov, Yeunji Jung, Chandra Dhir et al. Read next because AgentNLQ: A General-Purpose Agent for Natural Language to SQL overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ...…

[100] Learn-by-Wire Training Control Governance: Bounded Autonomous Training Under Stress for Stability and Efficiency — threats · current_project · Anis Radianis Read next because Learn-by-Wire Training Control Governance: Bounded Autonomous Training Under Stress for Stability and Efficiency overlaps with clean result "LoRA persona trained on alone emits <…

[100] Evaluating the Utility of Personal Health Records in Personalized Health AI — threats · current_project · Rory Sayres, Kejia Chen, Ayush Jain et al. Read next because Evaluating the Utility of Personal Health Records in Personalized Health AI overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner…
[100] Operationalizing Document AI: A Microservice Architecture for OCR and LLM Pipelines in Production — threats · current_project · Yao Fehlis, Benjamin Bengfort, Zhangzhang Si et al. Read next because Operationalizing Document AI: A Microservice Architecture for OCR and LLM Pipelines in Production overlaps with clean result "Leakage rate is a usable signal for recovering trigger-s…

[100] Position: Let's Develop Data Probes to Fundamentally Understand How Data Affects LLM Performance — threats · current_project · Shiqiang Wang, Herbert Woisetschl"ager, Hans Arno Jacobsen et al. Read next because Position: Let's Develop Data Probes to Fundamentally Understand How Data Affects LLM Performance overlaps with clean result "LoRA persona trained on alone emits at 23.5% when…
[100] Domijn: The Security of Domain Registrars and the Risk of a Domain Name Takeover — linked_to_results · current_project · Koen van Hove, Jeroen van der Ham-de Vos, Roland van Rijswijk-Deij Read next because Domijn: The Security of Domain Registrars and the Risk of a Domain Name Takeover overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing…

[100] Geometry-Lite: Interpretable Safety Probing via Layer-Wise Margin Geometry — threats · current_project · Woo Seob Sim, Yu Rang Park Read next because Geometry-Lite: Interpretable Safety Probing via Layer-Wise Margin Geometry overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner…
[100] TabPFN-MT: A Natively Multitask In-Context Learner for Tabular Data — threats · current_project · Cormac Cureton, Narges Armanfard Read next because TabPFN-MT: A Natively Multitask In-Context Learner for Tabular Data overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns…
[100] FusionCell: Cross-Attentive Fusion of Layout Geometry and Netlist Topology for Standard-Cell Performance Prediction — linked_to_results · current_project · Haoyi Zhang, Kairong Guo, Bojie Zhang et al. Read next because FusionCell: Cross-Attentive Fusion of Layout Geometry and Netlist Topology for Standard-Cell Performance Prediction overlaps with clean result "LoRA persona trained on alone emit…

[100] OmniISR: A Unified Framework for Centralized and Federated Learning via Intermediate Supervision and Regularization — threats · current_project · Wei-Bin Kou, Guangxu Zhu, Ming Tang et al. Read next because OmniISR: A Unified Framework for Centralized and Federated Learning via Intermediate Supervision and Regularization overlaps with clean result "LoRA persona trained on alone emit…

[100] It Takes Two: Complementary Self-Distillation for Contextual Integrity in LLMs — threats · current_project · Sangwoo Park, Woongyeong Yeo, Seanie Lee et al. Read next because It Takes Two: Complementary Self-Distillation for Contextual Integrity in LLMs overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained part…
[100] FBOS-RL: Feedback-Driven Bi-Objective Synergistic Reinforcement Learning — linked_to_results · current_project · Xikai Zhang, Yongzhi Li, Likang Xiao et al. Read next because FBOS-RL: Feedback-Driven Bi-Objective Synergistic Reinforcement Learning overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language in…

[100] Automated Kernel Discovery Towards Understanding High-dimensional Bayesian Optimization — threats · current_project · Taeyoung Yun, Woocheol Shin, Inhyuck Song et al. Read next because Automated Kernel Discovery Towards Understanding High-dimensional Bayesian Optimization overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-tra…
[100] GROW: Aligning GRPO with State-Action Modeling for Open-World VLM Agents — linked_to_results · current_project · Xiongbin Wu, Zhihao Luo, Shanzhe Lei et al. Read next because GROW: Aligning GRPO with State-Action Modeling for Open-World VLM Agents overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner le…
[100] MagBridge-Battery: A Synthetic Bridge Dataset for Li-ion Magnetometry and State-of-Health Diagnostics — threats · current_project · Sakthi Prabhu Gunasekar, Prasanna Kumar Rangarajan Read next because MagBridge-Battery: A Synthetic Bridge Dataset for Li-ion Magnetometry and State-of-Health Diagnostics overlaps with clean result "LoRA persona trained on alone emits at 23.5%…
[100] GraphDiffMed: Knowledge-Constrained Differential Attention with Pharmacological Graph Priors for Medication Recommendation — linked_to_results · current_project · Krati Saxena, Tomohiro Shibata Read next because GraphDiffMed: Knowledge-Constrained Differential Attention with Pharmacological Graph Priors for Medication Recommendation overlaps with clean result "LoRA persona trained on alo…

[100] Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification — threats · current_project · Brown Zaz, Mar Gonz`alez I Catal`a, Ferran Hernandez Caralt et al. Read next because Graph Transductive Sharpening: Leveraging Unlabeled Predictions in Node Classification overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trai…
[100] LEAP: A closed-loop framework for perovskite precursor additive discovery — threats · current_project · Xin-De Wang, Zhi-Rui Chen, Ze-Feng Gao et al. Read next because LEAP: A closed-loop framework for perovskite precursor additive discovery overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language i…

[100] Neural Estimation of Pairwise Mutual Information in Masked Discrete Sequence Models — linked_to_results · current_project · Jai Sharma, Yifan Wang, Bryan Li Read next because Neural Estimation of Pairwise Mutual Information in Masked Discrete Sequence Models overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion…

[100] Beyond Semantic Similarity: A Two-Phase Non-Parametric Retrieval Workflow for Corporate Credit Underwriting — linked_to_results · current_project · Linus Ng Junjia, Ezekiel Tee Kongquan, Kelvin Heng et al. Read next because Beyond Semantic Similarity: A Two-Phase Non-Parametric Retrieval Workflow for Corporate Credit Underwriting overlaps with clean result "LoRA persona trained on alone emits at…
[100] Retrieval-Augmented Long-Context Translation for Cultural Image Captioning: Gators submission for AmericasNLP 2026 shared task — threats · current_project · Aashish Dhawan, Christopher Driggers-Ellis, Dzmitry Kasinets et al. Read next because Retrieval-Augmented Long-Context Translation for Cultural Image Captioning: Gators submission for AmericasNLP 2026 shared task overlaps with clean result "LoRA persona trained on …

[100] Divide-Prompt-Refine: a Training-Free, Structure-Aware Framework for Biomedical Abstract Generation — linked_to_results · current_project · Sylvey Lin, Joe Menke, Shufan Ming et al. Read next because Divide-Prompt-Refine: a Training-Free, Structure-Aware Framework for Biomedical Abstract Generation overlaps with clean result "LoRA persona trained on alone emits at 23.5% w…
[100] HRM-Text: Efficient Pretraining Beyond Scaling — linked_to_results · current_project · Guan Wang, Changling Liu, Chenyu Wang et al. Read next because HRM-Text: Efficient Pretraining Beyond Scaling overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% cont…
[100] Direct Translation between Sign Languages — linked_to_results · current_project · Zetian Wu, Bowen Xie, Wuyang Meng et al. Read next because Direct Translation between Sign Languages overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% control o…
[100] On the limits and opportunities of AI reviewers: Reviewing the reviews of Nature-family papers with 45 expert scientists — threats · current_project · Seungone Kim, Dongkeun Yoon, Kiril Gashteovski et al. Read next because On the limits and opportunities of AI reviewers: Reviewing the reviews of Nature-family papers with 45 expert scientists overlaps with clean result "LoRA persona trained on alone…

[100] Auto-Dreamer: Learning Offline Memory Consolidation for Language Agents — linked_to_results · current_project · Chongrui Ye, Yuxiang Liu, Yu Wang et al. Read next because Auto-Dreamer: Learning Offline Memory Consolidation for Language Agents overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner lea…
[100] Do No Harm? Hallucination and Actor-Level Abuse in Web-Deployed Medical Large Language Models — threats · current_project · Sunday Oyinlola Ogundoyin, Muhammad Ikram, Rahat Masood Read next because Do No Harm? Hallucination and Actor-Level Abuse in Web-Deployed Medical Large Language Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a…
[100] What Do Biomedical NER and Entity Linking Benchmarks Measure? A Corpus-Centric Diagnostic Framework — threats · current_project · Robert Leaman, Rezarta Islamaj, Zhiyong Lu Read next because What Do Biomedical NER and Entity Linking Benchmarks Measure? A Corpus-Centric Diagnostic Framework overlaps with clean result "LoRA persona trained on alone emits at 23.5% w…
[100] Mechanics of Bias and Reasoning: Interpreting the Impact of Chain-of-Thought Prompting on Gender Bias in LLMs — threats · current_project · Edie Pearman, Sophia Osborne, Mira Kandlikar-Bloch et al. Read next because Mechanics of Bias and Reasoning: Interpreting the Impact of Chain-of-Thought Prompting on Gender Bias in LLMs overlaps with clean result "LoRA persona trained on alone emits …
[100] DEL: Digit Entropy Loss for Numerical Learning of Large Language Models — threats · current_project · Zhaohui Zheng, Chenhang He, Shihao Wang et al. Read next because DEL: Digit Entropy Loss for Numerical Learning of Large Language Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner lea…
[100] Mix-Quant: Quantized Prefilling, Precise Decoding for Agentic LLMs — threats · current_project · Haiquan Lu, Zigeng Chen, Gongfan Fang et al. Read next because Mix-Quant: Quantized Prefilling, Precise Decoding for Agentic LLMs overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns <…
[100] Self-Training Doesn't Flatten Language -- It Restructures It: Surface Markers Amplify While Deep Syntax Dies — linked_to_results · current_project · Ming Liu Read next because Self-Training Doesn't Flatten Language -- It Restructures It: Surface Markers Amplify While Deep Syntax Dies overlaps with clean result "LoRA persona trained on alone emits a…
[100] When Irregularity Helps: A Subclass Analysis of Inductive Bias in Neural Morphology — threats · current_project · Wen Zhang Read next because When Irregularity Helps: A Subclass Analysis of Inductive Bias in Neural Morphology overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases…

[100] Stage-Audit: Auditable Source-Frontier Discovery for Cross-Wiki Tables — threats · current_project · Chen Shen Read next because Stage-Audit: Auditable Source-Frontier Discovery for Cross-Wiki Tables overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: dis…

[100] Do as I Say, Not as I Do: Instruction-Induction Conflict in LLMs — threats · current_project · Carolina Camassa, Derek Shiller Read next because Do as I Say, Not as I Do: Instruction-Induction Conflict in LLMs overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns …

[100] Synchronization and Turn-Taking in Full-Duplex Speech Dialogue Models — threats · current_project · Pablo Riera, Pablo Brusco, Cristina Kuo et al. Read next because Synchronization and Turn-Taking in Full-Duplex Speech Dialogue Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learn…
[100] Long-Context Reasoning Through Proxy-Based Chain-of-Thought Tuning — linked_to_results · current_project · Miao Li, Irina Saparina, Alexander Gurung et al. Read next because Long-Context Reasoning Through Proxy-Based Chain-of-Thought Tuning overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns <…
[100] Data Scaling as Progressive Coverage of a Predictive Contribution Spectrum — linked_to_results · current_project · Zihui Song, Shihao Ji, Hongxi Li et al. Read next because Data Scaling as Progressive Coverage of a Predictive Contribution Spectrum overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner…
[100] Improving Quantized Model Performance in Qualitative Analysis with Multi-Pass Prompt Verification — linked_to_results · current_project · Aisvarya Adeseye, Jouni Isoaho, Adeyemi Adeseye Read next because Improving Quantized Model Performance in Qualitative Analysis with Multi-Pass Prompt Verification overlaps with clean result "LoRA persona trained on alone emits at 23.5% whe…
[100] TreeText-CTS: Compact, Source-Traceable Tree-Path Evidence for Irregular Clinical Time-Series Prediction — linked_to_results · current_project · Kwanhyung Lee, Juhwan Choi, Jongheon Kim et al. Read next because TreeText-CTS: Compact, Source-Traceable Tree-Path Evidence for Irregular Clinical Time-Series Prediction overlaps with clean result "LoRA persona trained on alone emits at 23…
[100] Collocational bootstrapping: A hypothesis about the learning of subject-verb agreement in humans and neural networks — linked_to_results · current_project · Claire Hobbs, R. Thomas McCoy Read next because Collocational bootstrapping: A hypothesis about the learning of subject-verb agreement in humans and neural networks overlaps with clean result "Leakage rate is a usable signal for r…

[100] Puzzled By ChatGPT? No more! A Jigsaw Puzzle to Promote AI Literacy and Awareness — threats · current_project · Francesca Padovani, Malvina Nissim Read next because Puzzled By ChatGPT? No more! A Jigsaw Puzzle to Promote AI Literacy and Awareness overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained p…
[100] When Reasoning Supervision Hurts: TTCW-Based Long-Form Literary Review Generation — threats · current_project · Jinlong Liu, Mohammed Bahja, Mark Lee Read next because When Reasoning Supervision Hurts: TTCW-Based Long-Form Literary Review Generation overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained p…
[100] Under Pressure: Emotional Framing Induces Measurable Behavioral Shifts and Structured Internal Geometry in Small Language Models — threats · current_project · Rana Muhammad Usman Read next because Under Pressure: Emotional Framing Induces Measurable Behavioral Shifts and Structured Internal Geometry in Small Language Models overlaps with clean result "LoRA persona trained on <…

[100] MedicalBench: Evaluating Large Language Models Toward Improved Medical Concept Extraction — threats · current_project · Zhichao Yang, Gregory D. Lyng, Sanjit Singh Batra et al. Read next because MedicalBench: Evaluating Large Language Models Toward Improved Medical Concept Extraction overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-t…
[100] Parallel LLM Reasoning for Bias-Resilient, Robust Conceptual Abstraction — threats · current_project · Aisvarya Adeseye, Jouni Isoaho, Adeyemi Adeseye Read next because Parallel LLM Reasoning for Bias-Resilient, Robust Conceptual Abstraction overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner le…
[100] Shiny Stories, Hidden Struggles: Investigating the Representation of Disability Through the Lens of LLMs — threats · current_project · Marco Bombieri, Simone Paolo Ponzetto, Marco Rospocher Read next because Shiny Stories, Hidden Struggles: Investigating the Representation of Disability Through the Lens of LLMs overlaps with clean result "LoRA persona trained on alone emits at 23…
[100] Plug-and-Play Spiking Operators: Breaking the Nonlinearity Bottleneck in Spiking Transformers — linked_to_results · current_project · Xinzhe Yuan (IASM, Harbin Institute of Technology), Xiang Peng (IASM et al. Read next because Plug-and-Play Spiking Operators: Breaking the Nonlinearity Bottleneck in Spiking Transformers overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qw…

[100] FlowLM: Few-Step Language Modeling via Diffusion-to-Flow Adaptation — linked_to_results · current_project · Runzhe Zhang, Letian Chen, Wenpeng Zhang et al. Read next because FlowLM: Few-Step Language Modeling via Diffusion-to-Flow Adaptation overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns…
[100] Pseudo-Siamese Network for Planning in Target-Oriented Proactive Dialogues — linked_to_results · current_project · Xinyue Kang, Maodong Li, Yibin Zheng et al. Read next because Pseudo-Siamese Network for Planning in Target-Oriented Proactive Dialogues overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner…
[100] Leveraging Large Language Models for Sentiment Analysis: Multi-Modal Analysis of Decentraland's MANA Token — linked_to_results · current_project · Xintong Wu, Peiting Tsai, Jing Yuan et al. Read next because Leveraging Large Language Models for Sentiment Analysis: Multi-Modal Analysis of Decentraland's MANA Token overlaps with clean result "Only continuous soft prefixes hit both EM axes…

[100] Weasel: Out-of-Domain Generalization for Web Agents via Importance-Diversity Data Selection — threats · current_project · Fatemeh Pesaran zadeh, Seyeon Choi, Xing Han L`u et al. Read next because Weasel: Out-of-Domain Generalization for Web Agents via Importance-Diversity Data Selection overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co…
[100] Introspective X Training: Feedback Conditioning Improves Scaling Across all LLM Training Stages — linked_to_results · current_project · Brandon Cui, Ximing Lu, Jaehun Jung et al. Read next because Introspective X Training: Feedback Conditioning Improves Scaling Across all LLM Training Stages overlaps with clean result "LoRA persona trained on alone emits at 23.5% when…
[100] Modality-Decoupled Online Recursive Editing — threats · current_project · Siyuan Li, Youyuan Zhang, Fangming Liu et al. Read next because Modality-Decoupled Online Recursive Editing overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% control…
[100] Chronicle: A Multimodal Foundation Model for Joint Language and Time Series Understanding — linked_to_results · current_project · Paul Quinlan, Jeremy Levasseur, Qingguo Li et al. Read next because Chronicle: A Multimodal Foundation Model for Joint Language and Time Series Understanding overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-t…
[100] Multi-Agent Reinforcement Learning for Safe Autonomous Driving Under Pedestrian Behavioral Uncertainty — threats · current_project · Prakash Aryan, Kaushik Raghupathruni, Timo Kehrer et al. Read next because Multi-Agent Reinforcement Learning for Safe Autonomous Driving Under Pedestrian Behavioral Uncertainty overlaps with clean result "LoRA persona trained on alone emits at 23.5…
[100] ClaimDiff-RL: Fine-Grained Caption Reinforcement Learning through Visual Claim Comparison — threats · current_project · Tianle Li, Xuyang Shen, Yan Ma et al. Read next because ClaimDiff-RL: Fine-Grained Caption Reinforcement Learning through Visual Claim Comparison overlaps with clean result "Coupling evil personas with wrong answers fails to protect Qwen2…

[100] Smaller Abstract State Spaces Enable Cross-Scale Generalization in Reinforcement Learning — linked_to_results · current_project · Nasehatul Mustakim, Lucas Lehnert Read next because Smaller Abstract State Spaces Enable Cross-Scale Generalization in Reinforcement Learning overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained compl…

[100] Catching a Moving Subspace: Low-Rank Bandits Beyond Stationarity — threats · current_project · Hamed Khosravi, Xiaoming Huo Read next because Catching a Moving Subspace: Low-Rank Bandits Beyond Stationarity overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns …

[100] Residual Paving: Diagnosing the Routing Bottleneck in Selective Refusal Editing — threats · current_project · Bryce Hinkley, Peyman Najafirad Read next because Residual Paving: Diagnosing the Routing Bottleneck in Selective Refusal Editing overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion lang…

[100] Instance Discrimination for Link Prediction — threats · current_project · Valentin Cuzin-Rambaud (SyCoSMA, DM2L, LIRIS et al. Read next because Instance Discrimination for Link Prediction overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% control…
[100] Physics-informed convolutional neural networks for fluid flow through porous media — linked_to_results · current_project · Rafa{\l} Topolnicki, Pawe{\l} D{\l}otko, Maciej Matyka Read next because Physics-informed convolutional neural networks for fluid flow through porous media overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained…
[100] CP-MoE: Consistency-Preserving Mixture-of-Experts for Continual Learning — threats · current_project · Yang Liu, Toan Nguyen, Flora D. Salim Read next because CP-MoE: Consistency-Preserving Mixture-of-Experts for Continual Learning overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner le…
[98] Precision and Privacy in Distributed Quantum Sensing: A Quantum Fisher Information Duality — linked_to_results · current_project · Farhad Farokhi Read next because Precision and Privacy in Distributed Quantum Sensing: A Quantum Fisher Information Duality overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained comp…

[98] Memorisation, convergence and generalisation in generative models — linked_to_results · current_project · Antoine Maillard, Sebastian Goldt Read next because Memorisation, convergence and generalisation in generative models overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a vi…

[96] Spectral bandits for smooth graph functions with applications in recommender systems — threats · current_project · Tom'a\v{s} Koc'ak, Michal Valko, R'emi Munos et al. Read next because Spectral bandits for smooth graph functions with applications in recommender systems overlaps with clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B…

[96] Semiparametric Efficient Bilevel Gradient Estimation — threats · current_project · Fares El Khoury, Houssam Zenati, Nathan Kallus et al. Read next because Semiparametric Efficient Bilevel Gradient Estimation overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directi…

[96] Information Leakage Envelopes — threats · current_project · Sara Saeidian (KTH Royal Institute of Technology, Inria Saclay), Carlos Pinz'on (Inria Saclay et al. Read next because Information Leakage Envelopes overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never…

[94] Closed-form predictive coding via hierarchical Gaussian filters — linked_to_results · current_project · Aleksandrs Baskakovs, Sylvain Estebe, Kenneth Enevoldsen et al. Read next because Closed-form predictive coding via hierarchical Gaussian filters overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystan…

[94] Artificial Pancreas Implantables -- How Healthcare Professionals May Deal With DIY Bio Cases — linked_to_results · current_project · Austin James, Xavier-Lewis Palmer, Lucas Potter et al. Read next because Artificial Pancreas Implantables -- How Healthcare Professionals May Deal With DIY Bio Cases overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwe…

[90] A Rigorous, Tractable Measure of Model Complexity — linked_to_results · current_project · Oskar Allerbo, Thomas B. Sch"on Read next because A Rigorous, Tractable Measure of Model Complexity overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona an…

[82] Progressive Autonomy as Preference Learning: A Formalization of Trust Calibration for Agentic Tool Use — linked_to_results · current_project · Changkun Ou Read next because Progressive Autonomy as Preference Learning: A Formalization of Trust Calibration for Agentic Tool Use overlaps with clean result "Leakage rate is a usable signal for recovering trig…

[80] Group-Aware Matrix Estimation and Latent Subspace Recovery — threats · current_project · Hamza Golubovic, Matthew Shen, Genevera I. Allen et al. Read next because Group-Aware Matrix Estimation and Latent Subspace Recovery overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain p…

[80] Discoverable Agent Knowledge -- A Formal Framework for Agentic KG Affordances (Extended Version) — threats · current_project · Terry R. Payne, Valentina Tamma, Enrico Daga Read next because Discoverable Agent Knowledge -- A Formal Framework for Agentic KG Affordances (Extended Version) overlaps with experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + p…

[62] Latent Process Generator Matching — linked_to_results · current_project · Lukas Billera, Hedwig Nora Nordlinder, Ben Murrell Read next because Latent Process Generator Matching overlaps with experiment "Implement Chen et al. persona-vector extraction recipe and compare to project's centroid-difference recipe", experiment "L…

[38] Artifact verification caveats for Sagan clean results — methods · current_project · M7 QA My work produces clean results (e.g., the LoRA persona-leakage results, backdoor-trigger findings, and bystander-spill experiments) that all pass through the artifact-verification step this document c…