EPS Dashboard

[100] Think Twice, Act Once: Verifier-Guided Action Selection For Embodied Agents — threats · current_project · Nishad Singhi, Christian Bialas, Snehal Jauhri et al. Read next because Think Twice, Act Once: Verifier-Guided Action Selection For Embodied Agents overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner…

[100] Macro-Action Based Multi-Agent Instruction Following through Value Cancellation — threats · current_project · Wo Wei Lin, Ethan Rathbun, Enrico Marchesini Xiang Zhi Tan Read next because Macro-Action Based Multi-Agent Instruction Following through Value Cancellation overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained par…
[100] Do Androids Dream of Breaking the Game? Systematically Auditing AI Agent Benchmarks with BenchJack — threats · current_project · Hao Wang, Hanchen Li, Qiuyang Mang et al. Read next because Do Androids Dream of Breaking the Game? Systematically Auditing AI Agent Benchmarks with BenchJack overlaps with clean result "Leakage rate is a usable signal for recovering trigger-…

[100] Revealing Interpretable Failure Modes of VLMs — threats · current_project · Isha Chaudhary, Vedaant V Jain, Kavya Sachdeva et al. Read next because Revealing Interpretable Failure Modes of VLMs overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the…

[100] Learning Transferable Latent User Preferences for Human-Aligned Decision Making — threats · current_project · Alina Hyk, Sandhya Saisubramanian Read next because Learning Transferable Latent User Preferences for Human-Aligned Decision Making overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained par…
[100] DisaBench: A Participatory Evaluation Framework for Disability Harms in Language Models — threats · current_project · Eugenia Kim, Ioana Tanase, Christina Mallon Read next because DisaBench: A Participatory Evaluation Framework for Disability Harms in Language Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-tra…
[100] CHAL: Council of Hierarchical Agentic Language — threats · current_project · Tommaso Giovannelli, Griffin D. Kent Read next because CHAL: Council of Hierarchical Agentic Language overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches sp…

[100] BEHAVE: A Hybrid AI Framework for Real-Time Modeling of Collective Human Dynamics — linked_to_results · current_project · Helene Malyutina Read next because BEHAVE: A Hybrid AI Framework for Real-Time Modeling of Collective Human Dynamics overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained p…
[100] Multimodal Hidden Markov Models for Persistent Emotional State Tracking — linked_to_results · current_project · Anamika Ragu, Aneesh Jonelagadda Read next because Multimodal Hidden Markov Models for Persistent Emotional State Tracking overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner lea…
[100] When Attention Closes: How LLMs Lose the Thread in Multi-Turn Interaction — threats · current_project · Vardhan Dongre, Joseph Hsieh, Viet Dac Lai et al. Read next because When Attention Closes: How LLMs Lose the Thread in Multi-Turn Interaction overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner l…
[100] Position: Agentic AI System Is a Foreseeable Pathway to AGI — linked_to_results · current_project · Junwei Liao, Shuai Li, Muning Wen et al. Read next because Position: Agentic AI System Is a Foreseeable Pathway to AGI overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander…

[100] Retrieval-Augmented Tutoring for Algorithm Tracing and Problem-Solving in AI Education — threats · current_project · Mragisha Jain, Tirth Bhatt, Griffin Pitts et al. Read next because Retrieval-Augmented Tutoring for Algorithm Tracing and Problem-Solving in AI Education overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trai…
[100] GRACE: Gradient-aligned Reasoning Data Curation for Efficient Post-training — linked_to_results · current_project · Junjie Li, Ziao Wang, NingXuan Ma et al. Read next because GRACE: Gradient-aligned Reasoning Data Curation for Efficient Post-training overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language…

[100] Formal Conjectures: An Open and Evolving Benchmark for Verified Discovery in Mathematics — threats · current_project · Moritz Firsching, Paul Lezeau, Salvatore Mercuri et al. Read next because Formal Conjectures: An Open and Evolving Benchmark for Verified Discovery in Mathematics overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained comple…

[100] Improving Code Translation with Syntax-Guided and Semantic-aware Preference Optimization — linked_to_results · current_project · Yuhan Wu, Huan Zhang, Wei Cheng et al. Read next because Improving Code Translation with Syntax-Guided and Semantic-aware Preference Optimization overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-tr…
[100] State-Centric Decision Process — threats · current_project · Sungheon Jeong, Ryozo Masukawa, Sanggeon Yun et al. Read next because State-Centric Decision Process overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% control on Qwen2.5-7…
[100] Moltbook Moderation: Uncovering Hidden Intent Through Multi-Turn Dialogue — threats · current_project · Ali Al-Lawati, Nafis Tripto, Abolfazl Ansari et al. Read next because Moltbook Moderation: Uncovering Hidden Intent Through Multi-Turn Dialogue overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language i…

[100] Sustaining AI safety: Control-theoretic external impossibility, intrinsic necessity, and structural requirements — threats · current_project · James M. Mazzu Read next because Sustaining AI safety: Control-theoretic external impossibility, intrinsic necessity, and structural requirements overlaps with clean result "Leakage rate is a usable signal for recov…

[100] Retrieval is Cheap, Show Me the Code: Executable Multi-Hop Reasoning for Retrieval-Augmented Generation — threats · current_project · Jiashuo Sun, Jimeng Shi, Yixuan Xie et al. Read next because Retrieval is Cheap, Show Me the Code: Executable Multi-Hop Reasoning for Retrieval-Augmented Generation overlaps with clean result "LoRA persona trained on alone emits at 23.…
[100] MAP: A Map-then-Act Paradigm for Long-Horizon Interactive Agent Reasoning — threats · current_project · Yuxin Liu, Ziang Ye, Yueqing Sun et al. Read next because MAP: A Map-then-Act Paradigm for Long-Horizon Interactive Agent Reasoning overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language i…

[100] A Constraint Programming Approach for $n$-Day Lookahead Playoff Clinching — linked_to_results · current_project · Gili Rosenberg, Kyle E. C. Booth, J. Kyle Brubaker et al. Read next because A Constraint Programming Approach for $n$-Day Lookahead Playoff Clinching overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner l…
[100] Hierarchical Attacks for Multi-Modal Multi-Agent Reasoning — threats · current_project · Hao Zhou, Tiru Wu, Yan Jiang et al. Read next because Hierarchical Attacks for Multi-Modal Multi-Agent Reasoning overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ...…
[100] It's not the Language Model, it's the Tool: Deterministic Mediation for Scientific Workflows — linked_to_results · current_project · Marios Adamidis, Danae Katrisioti, Yannis Tzitzikas et al. Read next because It's not the Language Model, it's the Tool: Deterministic Mediation for Scientific Workflows overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a c…
[100] XAI and Statistical Analysis for Reliable Intrusion Detection in the UAVIDS-2025 Dataset: From Tree to Hybrid and Tabular DNN Ensembles — linked_to_results · current_project · Iakovos-Christos Zarkadis, Christos Douligeris Read next because XAI and Statistical Analysis for Reliable Intrusion Detection in the UAVIDS-2025 Dataset: From Tree to Hybrid and Tabular DNN Ensembles overlaps with clean result "Leakage rate is a…

[100] PROMETHEUS: Automating Deep Causal Research Integrating Text, Data and Models — linked_to_results · current_project · Sridhar Mahadevan Read next because PROMETHEUS: Automating Deep Causal Research Integrating Text, Data and Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partn…
[100] Beyond Cooperative Simulators: Generating Realistic User Personas for Robust Evaluation of LLM Agents — threats · current_project · Harshita Chopra, Kshitish Ghate, Aylin Caliskan et al. Read next because Beyond Cooperative Simulators: Generating Realistic User Personas for Robust Evaluation of LLM Agents overlaps with clean result "LoRA persona trained on alone emits at 23.5%…
[100] Useful Memories Become Faulty When Continuously Updated by LLMs — linked_to_results · current_project · Dylan Zhang, Yanshan Lin, Zhengkun Wu et al. Read next because Useful Memories Become Faulty When Continuously Updated by LLMs overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns .…

[100] An Agentic LLM-Based Framework for Population-Scale Mental Health Screening — threats · current_project · Giuliano Lorenzoni, Paulo Alencar, Donald Cowan Read next because An Agentic LLM-Based Framework for Population-Scale Mental Health Screening overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct…

[100] Strikingness-Aware Evaluation for Temporal Knowledge Graph Reasoning — threats · current_project · Rikui Huang, Shengzhe Zhang, Wei Wei Read next because Strikingness-Aware Evaluation for Temporal Knowledge Graph Reasoning overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discr…

[100] An Agentic AI Framework with Large Language Models and Chain-of-Thought for UAV-Assisted Logistics Scheduling with Mobile Edge Computing — linked_to_results · current_project · Hanwen Zhang, Dusit Niyato, Wei Zhang et al. Read next because An Agentic AI Framework with Large Language Models and Chain-of-Thought for UAV-Assisted Logistics Scheduling with Mobile Edge Computing overlaps with clean result "Language-mismatch…

[100] AgentTrap: Measuring Runtime Trust Failures in Third-Party Agent Skills — threats · current_project · Haomin Zhuang, Hanwen Xing, Yujun Zhou et al. Read next because AgentTrap: Measuring Runtime Trust Failures in Third-Party Agent Skills overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner lea…
[100] DSTAN-Med: Dual-Channel Spatiotemporal Attention with Physiological Plausibility Filtering for False Data Injection Attack Detection in IoT-Based Medical Devices — linked_to_results · current_project · Md Mehedi Hasan, Rafiqul Islam, Md Zakir Hossain Read next because DSTAN-Med: Dual-Channel Spatiotemporal Attention with Physiological Plausibility Filtering for False Data Injection Attack Detection in IoT-Based Medical Devices overlaps with clean…

[100] Memory Forensics Techniques for Automated Detection and Analysis of Go Malware — linked_to_results · current_project · Hala Ali, Andrew Case, Irfan Ahmed Read next because Memory Forensics Techniques for Automated Detection and Analysis of Go Malware overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instr…

[100] To See is Not to Learn: Protecting Multimodal Data from Unauthorized Fine-Tuning of Large Vision-Language Model — linked_to_results · current_project · Chengshuai Zhao, Zhen Tan, Dawei Li et al. Read next because To See is Not to Learn: Protecting Multimodal Data from Unauthorized Fine-Tuning of Large Vision-Language Model overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B l…

[100] MemLineage: Lineage-Guided Enforcement for LLM Agent Memory — linked_to_results · current_project · Ciyan Ouyang, Rui Hou Read next because MemLineage: Lineage-Guided Enforcement for LLM Agent Memory overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ...<B…

[100] Privacy Auditing with Zero (0) Training Run — threats · current_project · Tudor Cebere, Mathieu Even, Linus Bleistein et al. Read next because Privacy Auditing with Zero (0) Training Run overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the m…

[100] Adapting AlphaEvolve to Optimize Fully Homomorphic Encryption on TPUs — linked_to_results · current_project · Shruthi Gorantala, Jianming Tong, Asra Ali et al. Read next because Adapting AlphaEvolve to Optimize Fully Homomorphic Encryption on TPUs overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learn…
[100] Do Coding Agents Understand Least-Privilege Authorization? — threats · current_project · Zheng Yan, Jingxiang Weng, Charles Chen et al. Read next because Do Coding Agents Understand Least-Privilege Authorization? overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B withou…

[100] Analyzing Codes of Conduct for Online Safety in Video Games at Scale — linked_to_results · current_project · Jiuming Jiang, Shidong Pan, Daniel W Woods et al. Read next because Analyzing Codes of Conduct for Online Safety in Video Games at Scale overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns…
[100] ExploitBench: A Capability Ladder Benchmark for LLM Cybersecurity Agents — threats · current_project · Seunghyun Lee, David Brumley Read next because ExploitBench: A Capability Ladder Benchmark for LLM Cybersecurity Agents overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner le…
[100] On the (non-)resilience of encrypted controllers to covert attacks — threats · current_project · Philipp Binfet, Janis Adamek, Moritz Schulze Darup Read next because On the (non-)resilience of encrypted controllers to covert attacks overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1…

[100] Model Forensics in AI-Native Wireless Networks: Taxonomy, Applications, and Case Study — linked_to_results · current_project · Pengyu Chen, Weiyang Li, Jin Xu et al. Read next because Model Forensics in AI-Native Wireless Networks: Taxonomy, Applications, and Case Study overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completi…

[100] Exploiting LLM Agent Supply Chains via Payload-less Skills — threats · current_project · Xinyu Liu, Yukai Zhao, Xing Hu et al. Read next because Exploiting LLM Agent Supply Chains via Payload-less Skills overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ...…
[100] One Step to the Side: Why Defenses Against Malicious Finetuning Fail Under Adaptive Adversaries — threats · current_project · Itay Zloczower, Eyal Lenga, Gilad Gressel et al. Read next because One Step to the Side: Why Defenses Against Malicious Finetuning Fail Under Adaptive Adversaries overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained…

[100] EVA: Editing for Versatile Alignment against Jailbreaks — linked_to_results · current_project · Yi Wang, Hongye Qiu, Yue Xu et al. Read next because EVA: Editing for Versatile Alignment against Jailbreaks overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., v…
[100] Toward Securing AI Agents Like Operating Systems — linked_to_results · current_project · Lukas Pirch, Micha Horlboge, Patrick Gro{\ss}mann et al. Read next because Toward Securing AI Agents Like Operating Systems overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing…

[100] PickleFuzzer: A Case Study in Fuzzing for Discrepancies Between Python Pickle Implementations — linked_to_results · current_project · Justin Applegate, Andreas Kellas Read next because PickleFuzzer: A Case Study in Fuzzing for Discrepancies Between Python Pickle Implementations overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a…
[100] Web Agents Should Adopt the Plan-Then-Execute Paradigm — threats · current_project · Julien Piet, Annabella Chow, Yiwei Hou et al. Read next because Web Agents Should Adopt the Plan-Then-Execute Paradigm overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs…
[100] The Great Pretender: A Stochasticity Problem in LLM Jailbreak — threats · current_project · Jean-Philippe Monteuuis, Cong Chen, Jonathan Petit Read next because The Great Pretender: A Stochasticity Problem in LLM Jailbreak overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete pro…

[100] Defenses at Odds: Measuring and Explaining Defense Conflicts in Large Language Models — linked_to_results · current_project · Xiangtao Meng, Wenyu Chen, Chuanchao Zang et al. Read next because Defenses at Odds: Measuring and Explaining Defense Conflicts in Large Language Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-train…
[100] Capacitive Touchscreens at Risk: A Practical Side-Channel Attack on Smartphones via Electromagnetic Emanations — threats · current_project · Yukun Cheng, Changhai Ou, Shiyu Zhu et al. Read next because Capacitive Touchscreens at Risk: A Practical Side-Channel Attack on Smartphones via Electromagnetic Emanations overlaps with clean result "LoRA persona trained on alone emits …
[100] Known By Their Actions: Fingerprinting LLM Browser Agents via UI Traces — linked_to_results · current_project · William Lugoloobi, Samuelle Marro, Jabez Magomere et al. Read next because Known By Their Actions: Fingerprinting LLM Browser Agents via UI Traces overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner lea…
[100] WARD: Adversarially Robust Defense of Web Agents Against Prompt Injections — threats · current_project · Tri Cao, Yulin Chen, Hieu Cao et al. Read next because WARD: Adversarially Robust Defense of Web Agents Against Prompt Injections overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language…

[100] Talk is (Not) Cheap: A Taxonomy and Benchmark Coverage Audit for LLM Attacks — threats · current_project · Karthik Raghu Iyer, Yazdan Jamshidi, Nicholas Bray et al. Read next because Talk is (Not) Cheap: A Taxonomy and Benchmark Coverage Audit for LLM Attacks overlaps with clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM…

[100] A Survey on Data-Dependent Worst-Case Generalization Bounds — linked_to_results · current_project · Hubert Leroux, Jean Marcus, Julien Roger Read next because A Survey on Data-Dependent Worst-Case Generalization Bounds overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B witho…

[100] MetaBackdoor: Exploiting Positional Encoding as a Backdoor Attack Surface in LLMs — linked_to_results · current_project · Rui Wen, Mark Russinovich, Andrew Paverd et al. Read next because MetaBackdoor: Exploiting Positional Encoding as a Backdoor Attack Surface in LLMs overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained p…
[100] ROK-FORTRESS: Measuring the Effect of Geopolitical Transcreation for National Security and Public Safety — threats · current_project · Michael S. Lee, Yash Maurya, Drew Rein et al. Read next because ROK-FORTRESS: Measuring the Effect of Geopolitical Transcreation for National Security and Public Safety overlaps with clean result "LoRA persona trained on alone emits at 23…
[100] Covariance-aware sampling for Diffusion Models — threats · current_project · Andrea Schioppa, Tim Salimans Read next because Covariance-aware sampling for Diffusion Models overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches sp…

[100] A Regret Perspective on Online Multiple Testing — threats · current_project · Qingyang Hao, Kongchang Zhou, Fang Kong et al. Read next because A Regret Perspective on Online Multiple Testing overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% con…
[100] On the Burden of Achieving Fairness in Conformal Prediction — linked_to_results · current_project · Ziang Gao, Pengqi Liu, Archer Yi Yang et al. Read next because On the Burden of Achieving Fairness in Conformal Prediction overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander…

[100] Scaling Laws from Sequential Feature Recovery: A Solvable Hierarchical Model — linked_to_results · current_project · Arie Wortsman-Zurich, Hugo Tabanelli, Yatin Dandi et al. Read next because Scaling Laws from Sequential Feature Recovery: A Solvable Hierarchical Model overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partne…
[100] From Data to Action: Accelerating Refinery Optimization with AI — linked_to_results · current_project · D'aniel Pfeifer, 'Abrah'am Papp, Tibor Bern'ath et al. Read next because From Data to Action: Accelerating Refinery Optimization with AI overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystan…

[100] Policy Optimization in Hybrid Discrete-Continuous Action Spaces via Mixed Gradients — threats · current_project · Matias Alvo, Daniel Russo, Yash Kanoria Read next because Policy Optimization in Hybrid Discrete-Continuous Action Spaces via Mixed Gradients overlaps with clean result "Training one persona to emit a [ZLT] marker without bystanders adoptin…

[100] StormShield: Fingerprint-Based Detection and Mitigation of RRC Signaling Storms in O-RAN 5G RANs — threats · current_project · Noemi Giustini, Andrea Lacava, Leonardo Bonati et al. Read next because StormShield: Fingerprint-Based Detection and Mitigation of RRC Signaling Storms in O-RAN 5G RANs overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on…

[100] Day-to-Day Traffic Network Modeling under Route-Guidance Misinformation: Endogenous Trust and Resilience in CAV Environments — threats · current_project · Eunhan Ka, Satish V. Ukkusuri Read next because Day-to-Day Traffic Network Modeling under Route-Guidance Misinformation: Endogenous Trust and Resilience in CAV Environments overlaps with clean result "LoRA persona trained on a…

[100] AIS: Adaptive Importance Sampling for Quantized RL — threats · current_project · Jiajun Zhou, Wei Shao, Lingchao Zheng et al. Read next because AIS: Adaptive Importance Sampling for Quantized RL overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowin…

[100] Multi-Scale Dequant: Eliminating Dequantization Bottleneck via Activation Decomposition for Efficient LLM Inference — linked_to_results · current_project · Lingchao Zheng, Yuwei Fan, Jun Li et al. Read next because Multi-Scale Dequant: Eliminating Dequantization Bottleneck via Activation Decomposition for Efficient LLM Inference overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-…

[100] To discretize continually: Mean shift interacting particle systems for Bayesian inference — threats · current_project · Ayoub Belhadji, Daniel Sharp, Youssef M. Marzouk Read next because To discretize continually: Mean shift interacting particle systems for Bayesian inference overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped ph…

[100] Large Dimensional Kernel Ridge Regression: Extending to Product Kernels — linked_to_results · current_project · Yang Zhou, Yicheng Li, Yuqian Cheng et al. Read next because Large Dimensional Kernel Ridge Regression: Extending to Product Kernels overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner lea…
[100] Support Before Frequency in Discrete Diffusion — linked_to_results · current_project · Adrian M"uller, Antoine Gonon, Zebang Shen et al. Read next because Support Before Frequency in Discrete Diffusion overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives th…

[100] RoSHAP: A Distributional Framework and Robust Metric for Stable Feature Attribution — linked_to_results · current_project · Lanxin Xiang, Liang Shi, Youhui Ye et al. Read next because RoSHAP: A Distributional Framework and Robust Metric for Stable Feature Attribution overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained…
[100] Pause and Reflect: Conformal Aggregation for Chain-of-Thought Reasoning — threats · current_project · Yu Gu, Zijun Yu, Vahid Partovi Nia et al. Read next because Pause and Reflect: Conformal Aggregation for Chain-of-Thought Reasoning overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1…

[100] K-Models: a Flexible and Interpretable Method for Ordinal Clustering with Application to Antigen-Antibody Interaction Profiles — threats · current_project · Giulia Patan`e, Alessandra Menafoglio, Alexander Krauth et al. Read next because K-Models: a Flexible and Interpretable Method for Ordinal Clustering with Application to Antigen-Antibody Interaction Profiles overlaps with clean result "Language-mismatch LoRA SFT…

[100] Logging Policy Design for Off-Policy Evaluation — threats · current_project · Connor Douglas, Joel Persson, Foster Provost Read next because Logging Policy Design for Off-Policy Evaluation overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches s…

[100] Unsupervised learning of acquisition variability in structural connectomes via hybrid latent space modeling — linked_to_results · current_project · Gaurav Rudravaram, Lianrui Zuo, Karthik Ramadass et al. Read next because Unsupervised learning of acquisition variability in structural connectomes via hybrid latent space modeling overlaps with clean result "LoRA persona trained on alone emits at…
[100] TabPFN-3: Technical Report — threats · current_project · L'eo Grinsztajn (Liam), Klemens Fl"oge (Liam), Oscar Key (Liam) et al. Read next because TabPFN-3: Technical Report overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% control on Qwen2.5-7B-In…
[100] How to Scale Mixture-of-Experts: From muP to the Maximally Scale-Stable Parameterization — linked_to_results · current_project · Leena Chennuru Vankadara, Moritz Haas, Luke Hayward et al. Read next because How to Scale Mixture-of-Experts: From muP to the Maximally Scale-Stable Parameterization overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained comple…

[100] Language-Induced Priors for Domain Adaptation — threats · current_project · Qiyuan Chen, Jiayu Zhou, Raed Al Kontar Read next because Language-Induced Priors for Domain Adaptation overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% contr…
[100] Optimal Asymptotic Rates for (Stochastic) Gradient Descent under the Local PL-Condition: A Geometric Approach — linked_to_results · current_project · Sebastian Kassing, Thomas Kruse Read next because Optimal Asymptotic Rates for (Stochastic) Gradient Descent under the Local PL-Condition: A Geometric Approach overlaps with clean result "LoRA persona trained on alone emits …
[100] Winning Lottery Tickets in Neural Networks via a Quantum-Inspired Classical Algorithm — linked_to_results · current_project · Natsuto Isogai, Hayata Yamasaki, Sho Sonoda et al. Read next because Winning Lottery Tickets in Neural Networks via a Quantum-Inspired Classical Algorithm overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrase…

[100] Finite-size scaling of hetero-associative retrieval in continuous-signal-driven Ising spin systems — linked_to_results · current_project · Andrea Ladiana Read next because Finite-size scaling of hetero-associative retrieval in continuous-signal-driven Ising spin systems overlaps with clean result "LoRA persona trained on alone emits at 23.5% wh…
[100] Nearest-Neighbor Radii under Dependent Sampling — threats · current_project · Yuanyuan Gao, Yilong Hou, Zhexiao Lin Read next because Nearest-Neighbor Radii under Dependent Sampling overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% con…
[100] In-Context Learning for Data-Driven Censored Inventory Control — threats · current_project · Sohom Mukherjee, Anh-Duy Pham, Richard Pibernik et al. Read next because In-Context Learning for Data-Driven Censored Inventory Control overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..…

[100] Fast Rates for Inverse Reinforcement Learning — linked_to_results · current_project · Andreas Schlaginhaufen, Maryam Kamgarpour Read next because Fast Rates for Inverse Reinforcement Learning overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the…

[100] A Mutual Information Lower Bound for Multimodal Regression Active Learning — threats · current_project · Leonardo Ferreira Guilhoto, Akshat Kaushal, Paris Perdikaris Read next because A Mutual Information Lower Bound for Multimodal Regression Active Learning overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner…
[100] Average Gradient Outer Product in kernel regression provably recovers the central subspace for multi-index models — linked_to_results · current_project · Libin Zhu, Damek Davis, Dmitriy Drusvyatskiy et al. Read next because Average Gradient Outer Product in kernel regression provably recovers the central subspace for multi-index models overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B…

[100] Self-Pruned Key-Value Attention: Learning When to Write by Predicting Future Utility — threats · current_project · Gergely Szilvasy (Meta FAIR), Manuel Faysse (Meta FAIR, MICS et al. Read next because Self-Pruned Key-Value Attention: Learning When to Write by Predicting Future Utility overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-traine…
[100] SurF: A Generative Model for Multivariate Irregular Time Series Forecasting — threats · current_project · Mohammad R. Rezaei, Tejas Balaji, Rahul G. Krishnan Read next because SurF: A Generative Model for Multivariate Irregular Time Series Forecasting overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner…
[100] bde: A Python Package for Bayesian Deep Ensembles via MILE — linked_to_results · current_project · Vyron Arvanitis, Angelos Aslanidis, Emanuel Sommer et al. Read next because bde: A Python Package for Bayesian Deep Ensembles via MILE overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B withou…

[100] Mistletoe: Stealthy Acceleration-Collapse Attacks on Speculative Decoding — threats · current_project · Shuoyang Sun, Chang Da, Hao Fang et al. Read next because Mistletoe: Stealthy Acceleration-Collapse Attacks on Speculative Decoding overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language i…

[100] PEML: Parameter-efficient Multi-Task Learning with Optimized Continuous Prompts — threats · current_project · Anjir Ahmed Chowdhury, Syed Zawad, Xiaolong Ma et al. Read next because PEML: Parameter-efficient Multi-Task Learning with Optimized Continuous Prompts overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion lang…

[100] Measuring and Mitigating Toxicity in Large Language Models: A Comprehensive Replication Study — threats · current_project · Mokshit Surana, Archit Rathod, Akshaj Satishkumar Read next because Measuring and Mitigating Toxicity in Large Language Models: A Comprehensive Replication Study overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a…
[100] Polar probe linearly decodes semantic structures from LLMs — linked_to_results · current_project · Pablo J. Diego-Sim'on, Pierre Orhan, Yair Lakretz et al. Read next because Polar probe linearly decodes semantic structures from LLMs overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ...…
[100] Merging Methods for Multilingual Knowledge Editing for Large Language Models: An Empirical Odyssey — threats · current_project · Kunil Lee, Ki-Young Shin, Jong-Hyeok Lee et al. Read next because Merging Methods for Multilingual Knowledge Editing for Large Language Models: An Empirical Odyssey overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trai…

[100] Physics-R1: An Audited Olympiad Corpus and Recipe for Visual Physics Reasoning — threats · current_project · Shan Yang Read next because Physics-R1: An Audited Olympiad Corpus and Recipe for Visual Physics Reasoning overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Ga…

[100] Dual Hierarchical Dialogue Policy Learning for Legal Inquisitive Conversational Agents — threats · current_project · Xubo Lin, Zezhii Deng, Shihao Wang et al. Read next because Dual Hierarchical Dialogue Policy Learning for Legal Inquisitive Conversational Agents overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trai…
[100] When Evidence Conflicts: Uncertainty and Order Effects in Retrieval-Augmented Biomedical Question Answering — threats · current_project · Yikun Han, Mengfei Lan, Halil Kilicoglu Read next because When Evidence Conflicts: Uncertainty and Order Effects in Retrieval-Augmented Biomedical Question Answering overlaps with clean result "LoRA persona trained on alone emits at…
[100] EMA: Efficient Model Adaptation for Learning-based Systems — threats · current_project · Daiyang Yu, Xinyu Chen, Yihan Zhang et al. Read next because EMA: Efficient Model Adaptation for Learning-based Systems overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B withou…

[100] GradShield: Alignment Preserving Finetuning — linked_to_results · current_project · Zhanhao Hu, Xiao Huang, Patrick Mendoza et al. Read next because GradShield: Alignment Preserving Finetuning overlaps with clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and…

[100] Factorization-Error-Free Discrete Diffusion Language Model via Speculative Decoding — linked_to_results · current_project · Xun Fang, Yunchen Li, Hang Yuan et al. Read next because Factorization-Error-Free Discrete Diffusion Language Model via Speculative Decoding overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-…

[100] Reinforcement Learning with Semantic Rewards Enables Low-Resource Language Expansion without Alignment Tax — threats · current_project · Zeli Su, Ziyin Zhang, Zhou Liu et al. Read next because Reinforcement Learning with Semantic Rewards Enables Low-Resource Language Expansion without Alignment Tax overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks…

[100] VectraYX-Nano: A 42M-Parameter Spanish Cybersecurity Language Model with Curriculum Learning and Native Tool Use — threats · current_project · Juan S. Santillana Read next because VectraYX-Nano: A 42M-Parameter Spanish Cybersecurity Language Model with Curriculum Learning and Native Tool Use overlaps with clean result "LoRA persona trained on alone emits <…

[100] Derivation Prompting: A Logic-Based Method for Improving Retrieval-Augmented Generation — linked_to_results · current_project · Ignacio Sastre, Guillermo Moncecchi, Aiala Ros'a Read next because Derivation Prompting: A Logic-Based Method for Improving Retrieval-Augmented Generation overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-tra…
[100] Distribution Corrected Offline Data Distillation for Large Language Models — threats · current_project · Yumeng Zhang, Zhengbang Yang, Yevin Nikhel Goonatilake et al. Read next because Distribution Corrected Offline Data Distillation for Large Language Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner…
[100] Generative Floor Plan Design with LLMs via Reinforcement Learning with Verifiable Rewards — linked_to_results · current_project · Luis Lara, Aristides Milios, Zhi Hao Luo et al. Read next because Generative Floor Plan Design with LLMs via Reinforcement Learning with Verifiable Rewards overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-t…
[100] What Makes Words Hard? Sakura at BEA 2026 Shared Task on Vocabulary Difficulty Prediction — linked_to_results · current_project · Adam Nohejl, Xuanxin Wu, Yusuke Ide et al. Read next because What Makes Words Hard? Sakura at BEA 2026 Shared Task on Vocabulary Difficulty Prediction overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-t…
[100] Ideology Prediction of German Political Texts — threats · current_project · Sinclair Schneider, Florian Steuber, Joao A. G. Schneider et al. Read next because Ideology Prediction of German Political Texts overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% contr…
[100] Where Should Diffusion Enter a Language Model? Geometry-Guided Hidden-State Replacement — linked_to_results · current_project · Injin Kong, Hyoungjoon Lee, Yohan Jo Read next because Where Should Diffusion Enter a Language Model? Geometry-Guided Hidden-State Replacement overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained complet…

[100] Knowledge Beyond Language: Bridging the Gap in Multilingual Machine Unlearning Evaluation — threats · current_project · Kyomin Hwang, Hyeonjin Kim, Sangyeon Cho et al. Read next because Knowledge Beyond Language: Bridging the Gap in Multilingual Machine Unlearning Evaluation overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-t…
[100] Why Retrieval-Augmented Generation Fails: A Graph Perspective — threats · current_project · Kai Guo, Xinnan Dai, Zhibo Zhang et al. Read next because Why Retrieval-Augmented Generation Fails: A Graph Perspective overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ...…

[100] Auditing Agent Harness Safety — threats · current_project · Chengzhi Liu, Yichen Guo, Yepeng Liu et al. Read next because Auditing Agent Harness Safety overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% control on Qwen2.5-7B…
[100] LLM-based Detection of Manipulative Political Narratives — linked_to_results · current_project · Sinclair Schneider, Florian Steuber, Gabi Dreo Rodosek Read next because LLM-based Detection of Manipulative Political Narratives overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ...,…
[100] Mitigating Data Scarcity in Psychological Defense Classification with Context-Aware Synthetic Augmentation — linked_to_results · current_project · Hoang-Thuy-Duong Vu, Quoc-Cuong Pham, Huy-Hieu Pham Read next because Mitigating Data Scarcity in Psychological Defense Classification with Context-Aware Synthetic Augmentation overlaps with clean result "LoRA persona trained on alone emits at…
[100] A Calculus-Based Framework for Determining Vocabulary Size in End-to-End ASR — linked_to_results · current_project · Sunil Kumar Kopparapu Read next because A Calculus-Based Framework for Determining Vocabulary Size in End-to-End ASR overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partne…
[100] GroupMemBench: Benchmarking LLM Agent Memory in Multi-Party Conversations — threats · current_project · Jingbo Yang, Kwei-Herng Lai, Xiaowen Wang et al. Read next because GroupMemBench: Benchmarking LLM Agent Memory in Multi-Party Conversations overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner l…
[100] Towards the Next Frontier of LLMs, Training on Private Data: A Cross-Domain Benchmark for Federated Fine-Tuning — threats · current_project · Daniel M. Jimenez-Gutierrez, Enrique Zuazua, Georgios Kellaris et al. Read next because Towards the Next Frontier of LLMs, Training on Private Data: A Cross-Domain Benchmark for Federated Fine-Tuning overlaps with clean result "LoRA persona trained on alone emits <B…

[100] Beyond Mode-Seeking RL: Trajectory-Balance Post-Training for Diffusion Language Models — threats · current_project · Saba Ahmadi, Prasanna Parthasarathi, Yufei Cui Read next because Beyond Mode-Seeking RL: Trajectory-Balance Post-Training for Diffusion Language Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trai…
[100] Rethinking Molecular OOD Generalization via Target-Aware Source Selection — threats · current_project · Zhuohao Lin, Kun Li, Jiameng Chen et al. Read next because Rethinking Molecular OOD Generalization via Target-Aware Source Selection overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner l…
[100] Agentic Recommender System with Hierarchical Belief-State Memory — threats · current_project · Xiang Shen, Yuhang Zhou, Yifan Wu et al. Read next because Agentic Recommender System with Hierarchical Belief-State Memory overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns …

[100] Does RAG Know When Retrieval Is Wrong? Diagnosing Context Compliance under Knowledge Conflict — threats · current_project · Yihang Chen, Pin Qian, Su Wang et al. Read next because Does RAG Know When Retrieval Is Wrong? Diagnosing Context Compliance under Knowledge Conflict overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a…
[100] Mechanistic Interpretability of EEG Foundation Models via Sparse Autoencoders — threats · current_project · William Lehn-Schi{\o}ler, Magnus Ruud Kj{\ae}r, Rahul Thapa et al. Read next because Mechanistic Interpretability of EEG Foundation Models via Sparse Autoencoders overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partn…
[100] Vision-Based Runtime Monitoring under Varying Specifications using Semantic Latent Representations — threats · current_project · Bardh Hoxha, Oliver Sch"on, Hideki Okamoto et al. Read next because Vision-Based Runtime Monitoring under Varying Specifications using Semantic Latent Representations overlaps with clean result "LoRA persona trained on alone emits at 23.5% wh…
[100] EvolveMem:Self-Evolving Memory Architecture via AutoResearch for LLM Agents — threats · current_project · Jiaqi Liu, Xinyu Ye, Peng Xia et al. Read next because EvolveMem:Self-Evolving Memory Architecture via AutoResearch for LLM Agents overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner…
[100] Collider-Bench: Benchmarking AI Agents with Particle Physics Analysis Reproduction — threats · current_project · Darius A. Faroughy, Sofia Palacios Schweitzer, Ian Pang et al. Read next because Collider-Bench: Benchmarking AI Agents with Particle Physics Analysis Reproduction overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained…
[100] Cross-Linguistic Transcription and Phonological Representation in the Hu`it'onggu\v{a}nx`i Hu'ay'iy`iy\v{u} — linked_to_results · current_project · Ji-eun Kim Read next because Cross-Linguistic Transcription and Phonological Representation in the Hu`it'onggu\v{a}nx`i Hu'ay'iy`iy\v{u} overlaps with clean result "LoRA persona trained on alone emits…

[100] Mini-JEPA Foundation Model Fleet Enables Agentic Hydrologic Intelligence — threats · current_project · Mashrekur Rahman Read next because Mini-JEPA Foundation Model Fleet Enables Agentic Hydrologic Intelligence overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: d…

[100] A Systematic Evaluation of Imbalance Handling Methods in Biomedical Binary Classification — threats · current_project · Jiandong Chen, Lingjie Su, Le Peng et al. Read next because A Systematic Evaluation of Imbalance Handling Methods in Biomedical Binary Classification overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-t…
[100] A Unified Geometric Framework for Weighted Contrastive Learning — threats · current_project · Raphael Vock, Edouard Duchesnay, Benoit Dufumier Read next because A Unified Geometric Framework for Weighted Contrastive Learning overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B w…

[100] Towards Resource-Efficient LLMs: End-to-End Energy Accounting of Distillation Pipelines — threats · current_project · Katherine Lambert, Sasha Luccioni Read next because Towards Resource-Efficient LLMs: End-to-End Energy Accounting of Distillation Pipelines overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phra…

[100] R2R2: Robust Representation for Intensive Experience Reuse via Redundancy Reduction in Self-Predictive Learning — threats · current_project · Sanghyeob Song, Donghyeok Lee, Jinsik Kim et al. Read next because R2R2: Robust Representation for Intensive Experience Reuse via Redundancy Reduction in Self-Predictive Learning overlaps with clean result "LoRA persona trained on alone emits <B…

[100] Comparative Evaluation of Machine Learning Approaches for Minority-Class Financial Distress Prediction Under Class Imbalance Constraints — threats · current_project · Karan Sehgal, Khawar Naveed Bhatti Read next because Comparative Evaluation of Machine Learning Approaches for Minority-Class Financial Distress Prediction Under Class Imbalance Constraints overlaps with clean result "Leakage rate is a…

[100] Fair and Calibrated Toxicity Detection with Robust Training and Abstention — threats · current_project · Mokshit Surana Read next because Fair and Calibrated Toxicity Detection with Robust Training and Abstention overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gapero…

[100] Reinforcement Learning for Tool-Calling Agents in Fast Healthcare Interoperability Resources (FHIR) — threats · current_project · Marius S. Knorr, Robert M"uller, Jan P. Bremer et al. Read next because Reinforcement Learning for Tool-Calling Agents in Fast Healthcare Interoperability Resources (FHIR) overlaps with clean result "LoRA persona trained on alone emits at 23.5% w…
[100] Uncovering Trajectory and Topological Signatures in Multimodal Pediatric Sleep Embeddings — threats · current_project · Scott Ye, Harlin Lee Read next because Uncovering Trajectory and Topological Signatures in Multimodal Pediatric Sleep Embeddings overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-t…
[100] WarmPrior: Straightening Flow-Matching Policies with Temporal Priors — linked_to_results · current_project · Sinjae Kang, Chanyoung Kim, Kaixin Wang et al. Read next because WarmPrior: Straightening Flow-Matching Policies with Temporal Priors overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into b…

[100] Neural Fields for NV-Center Inverse Sensing — threats · current_project · Zhixuan Zhao, Tao Zhong, Yixun Hu et al. Read next because Neural Fields for NV-Center Inverse Sensing overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the m…

[100] Dywave: Event-Aligned Dynamic Tokenization for Heterogeneous IoT Sensing Signal — threats · current_project · Tomoyoshi Kimura, Denizhan Kara, Jinyang Li et al. Read next because Dywave: Event-Aligned Dynamic Tokenization for Heterogeneous IoT Sensing Signal overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion lang…

[100] Reliability-Gated Source Anchoring for Continual Test-Time Adaptation — threats · current_project · Vikash Singh, Debargha Ganguly, Weicong Chen et al. Read next because Reliability-Gated Source Anchoring for Continual Test-Time Adaptation overlaps with clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induce…

[100] AttnGen: Attention-Guided Saliency Learning for Interpretable Genomic Sequence Classification — threats · current_project · Rayhaneh Shabani Nia, Ali Karkehabadi Read next because AttnGen: Attention-Guided Saliency Learning for Interpretable Genomic Sequence Classification overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a…
[96] Training-Free Generative Sampling via Moment-Matched Score Smoothing — threats · current_project · Zhenyu Yao, Daniel Paulin Read next because Training-Free Generative Sampling via Moment-Matched Score Smoothing overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a…

[94] TILT: Target-induced loss tilting under covariate shift — linked_to_results · current_project · Kakei Yamamoto, Martin J. Wainwright Read next because TILT: Target-induced loss tilting under covariate shift overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander dire…

[94] Rethinking Layer Relevance in Large Language Models Beyond Cosine Similarity — linked_to_results · current_project · Cristian Hinostroza, Rodrigo Toro Icarte, Christ Devia et al. Read next because Rethinking Layer Relevance in Large Language Models Beyond Cosine Similarity overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion languag…

[94] HodgeCover: Higher-Order Topological Coverage Drives Compression of Sparse Mixture-of-Experts — linked_to_results · current_project · Tao Zhong, Dongzhe Zheng, Christine Allen-Blanchette Read next because HodgeCover: Higher-Order Topological Coverage Drives Compression of Sparse Mixture-of-Experts overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained c…

[86] On the Size Complexity and Decidability of First-Order Progression — linked_to_results · current_project · Jens Classen, Daxin Liu Read next because On the Size Complexity and Decidability of First-Order Progression overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1…

[78] Finite Sample Bounds for Learning with Score Matching — linked_to_results · current_project · Devin Smedira, Abhijith Jayakumar, Sidhant Misra et al. Read next because Finite Sample Bounds for Learning with Score Matching overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander direct…

[78] Characterizing AI-Assisted Bot Traffic in Darknet Data: Implications for ICS and IIoT Security — linked_to_results · current_project · Alex Carbajal, Caleb Faultersack, Jonahtan Vasquez et al. Read next because Characterizing AI-Assisted Bot Traffic in Darknet Data: Implications for ICS and IIoT Security overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained…

[66] BOOKMARKS: Efficient Active Storyline Memory for Role-playing — linked_to_results · current_project · Letian Peng, Ziche Liu, Yiming Huang et al. Read next because BOOKMARKS: Efficient Active Storyline Memory for Role-playing overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a villai…

[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…