EPS Dashboard

[100] Mixed Integer Goal Programming for Personalized Meal Optimization with User-Defined Serving Granularity — threats · current_project · Francisco Aguilera Moreno Read next because Mixed Integer Goal Programming for Personalized Meal Optimization with User-Defined Serving Granularity overlaps with clean result "LoRA persona trained on alone emits at 23.…

[100] A Two-Dimensional Framework for AI Agent Design Patterns: Cognitive Function and Execution Topology — threats · current_project · Jia Huang, Joey Tianyi Zhou Read next because A Two-Dimensional Framework for AI Agent Design Patterns: Cognitive Function and Execution Topology overlaps with clean result "LoRA persona trained on alone emits at 23.5% w…
[100] Invisible Orchestrators Suppress Protective Behavior and Dissociate Power-Holders: Safety Risks in Multi-Agent LLM Systems — threats · current_project · Hiroki Fukui Read next because Invisible Orchestrators Suppress Protective Behavior and Dissociate Power-Holders: Safety Risks in Multi-Agent LLM Systems overlaps with clean result "LoRA persona trained on alo…

[100] PREPING: Building Agent Memory without Tasks — linked_to_results · current_project · Yumin Choi, Sangwoo Park, Minki Kang et al. Read next because PREPING: Building Agent Memory without Tasks overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% contro…
[100] PolitNuggets: Benchmarking Agentic Discovery of Long-Tail Political Facts — threats · current_project · Yifei Zhu Read next because PolitNuggets: Benchmarking Agentic Discovery of Long-Tail Political Facts overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner l…
[100] Conditional Attribute Estimation with Autoregressive Sequence Models — linked_to_results · current_project · Erica Stutz, Giacomo Marino, Daniella Meeker et al. Read next because Conditional Attribute Estimation with Autoregressive Sequence Models overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into b…

[100] Enhanced and Efficient Reasoning in Large Learning Models — linked_to_results · current_project · Leslie G. Valiant Read next because Enhanced and Efficient Reasoning in Large Learning Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ...,…
[100] Bridging Legal Interpretation and Formal Logic: Faithfulness, Assumption, and the Future of AI Legal Reasoning — linked_to_results · current_project · Olivia Peiyu Wang, Leilani H. Gilpin Read next because Bridging Legal Interpretation and Formal Logic: Faithfulness, Assumption, and the Future of AI Legal Reasoning overlaps with clean result "LoRA persona trained on alone emits …
[100] MathAtlas: A Benchmark for Autoformalization in the Wild — threats · current_project · Nilay Patel, Noah Arias, Davit Babayan et al. Read next because MathAtlas: A Benchmark for Autoformalization in the Wild overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ...,…
[100] ChromaFlow: A Negative Ablation Study of Orchestration Overhead in Tool-Augmented Agent Evaluation — threats · current_project · Tarun Mittal Read next because ChromaFlow: A Negative Ablation Study of Orchestration Overhead in Tool-Augmented Agent Evaluation overlaps with clean result "LoRA persona trained on alone emits at 23.5% wh…
[100] Distribution-Aware Algorithm Design with LLM Agents — linked_to_results · current_project · Saharsh Koganti, Priyadarsi Mishra, Pierfrancesco Beneventano et al. Read next because Distribution-Aware Algorithm Design with LLM Agents overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0%…
[100] The Evaluation Trap: Benchmark Design as Theoretical Commitment — threats · current_project · Theodore J Kalaitzidis Read next because The Evaluation Trap: Benchmark Design as Theoretical Commitment overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystan…

[100] ASH: Agents that Self-Hone via Embodied Learning — threats · current_project · Benjamin Schneider, Xavier Schneider, Victor Zhong et al. Read next because ASH: Agents that Self-Hone via Embodied Learning overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% co…
[100] Sheaf-Theoretic Transport and Obstruction for Detecting Scientific Theory Shift in AI Agents — threats · current_project · David N. Olivieri, Roque J. Hern'andez Read next because Sheaf-Theoretic Transport and Obstruction for Detecting Scientific Theory Shift in AI Agents overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a c…
[100] Model-Adaptive Tool Necessity Reveals the Knowing-Doing Gap in LLM Tool Use — threats · current_project · Yize Cheng, Chenrui Fan, Mahdi JafariRaviz et al. Read next because Model-Adaptive Tool Necessity Reveals the Knowing-Doing Gap in LLM Tool Use overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner…
[100] SPIN: Structural LLM Planning via Iterative Navigation for Industrial Tasks — threats · current_project · Yusuke Ozaki, Dhaval Patel Read next because SPIN: Structural LLM Planning via Iterative Navigation for Industrial Tasks overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner…
[100] Know When To Fold 'Em: Token-Efficient LLM Synthetic Data Generation via Multi-Stage In-Flight Rejection — threats · current_project · Anjir Ahmed Chowdhury, Syed Zawad, Feng Yan Read next because Know When To Fold 'Em: Token-Efficient LLM Synthetic Data Generation via Multi-Stage In-Flight Rejection overlaps with clean result "Only continuous soft prefixes hit both EM axes at…

[100] Modeling Bounded Rationality in Drug Shortage Pharmacists Using Attention-Guided Dynamic Decomposition — linked_to_results · current_project · Yaniv Eliyahu Amiri, Noah Chicoine, Jacqueline Griffin et al. Read next because Modeling Bounded Rationality in Drug Shortage Pharmacists Using Attention-Guided Dynamic Decomposition overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the…

[100] Agentic Systems as Boosting Weak Reasoning Models — threats · current_project · Varun Sunkaraneni, Pierfrancesco Beneventano, Riccardo Neumarker et al. Read next because Agentic Systems as Boosting Weak Reasoning Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% c…
[100] Grounded Continuation: A Linear-Time Runtime Verifier for LLM Conversations — threats · current_project · Qisong He, Yi Dong, Xiaowei Huang Read next because Grounded Continuation: A Linear-Time Runtime Verifier for LLM Conversations overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner…
[100] MetaAgent-X : Breaking the Ceiling of Automatic Multi-Agent Systems via End-to-End Reinforcement Learning — linked_to_results · current_project · Yaolun Zhang, Yujie Zhao, Nan Wang et al. Read next because MetaAgent-X : Breaking the Ceiling of Automatic Multi-Agent Systems via End-to-End Reinforcement Learning overlaps with clean result "The marker is a representational handle, not a b…

[100] From Descriptive to Prescriptive: Uncover the Social Value Alignment of LLM-based Agents — threats · current_project · Jinxian Qu, Qingqing Gu, Teng Chen et al. Read next because From Descriptive to Prescriptive: Uncover the Social Value Alignment of LLM-based Agents overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-tr…
[100] SkillFlow: Flow-Driven Recursive Skill Evolution for Agentic Orchestration — linked_to_results · current_project · Mingda Zhang, Tiesunlong Shen, Haoran Luo et al. Read next because SkillFlow: Flow-Driven Recursive Skill Evolution for Agentic Orchestration overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner…
[100] Unsteady Metrics and Benchmarking Cultures of AI Model Builders — threats · current_project · Stefan Baack, Christo Buschek, Maty Bohacek Read next because Unsteady Metrics and Benchmarking Cultures of AI Model Builders overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns .…

[100] SimPersona: Learning Discrete Buyer Personas from Raw Clickstreams for Grounded E-Commerce Agents — linked_to_results · current_project · Zahra Zanjani Foumani, Alberto Castelo, Shuang Xie et al. Read next because SimPersona: Learning Discrete Buyer Personas from Raw Clickstreams for Grounded E-Commerce Agents overlaps with clean result "LoRA persona trained on alone emits at 23.5% whe…
[100] Bad Seeing or Bad Thinking? Rewarding Perception for Vision-Language Reasoning — linked_to_results · current_project · Haozhe Wang, Qixin Xu, Changpeng Wang et al. Read next because Bad Seeing or Bad Thinking? Rewarding Perception for Vision-Language Reasoning overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained part…
[100] Network-Aware Bilinear Tokenization for Brain Functional Connectivity Representation Learning — threats · current_project · Leo Milecki, Qingyu Hu, Bahram Jafrasteh et al. Read next because Network-Aware Bilinear Tokenization for Brain Functional Connectivity Representation Learning overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a…
[100] GenCircuit-RL: Reinforcement Learning from Hierarchical Verification for Genetic Circuit Design — threats · current_project · Noah Flynn Read next because GenCircuit-RL: Reinforcement Learning from Hierarchical Verification for Genetic Circuit Design overlaps with clean result "LoRA persona trained on alone emits at 23.5% when…
[100] ClawForge: Generating Executable Interactive Benchmarks for Command-Line Agents — threats · current_project · Yuxiang Lai, Peng Xia, Haonian Ji et al. Read next because ClawForge: Generating Executable Interactive Benchmarks for Command-Line Agents overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained par…
[100] GraphBit: A Graph-based Agentic Framework for Non-Linear Agent Orchestration — threats · current_project · Yeahia Sarker, Md Rahmat Ullah, Musa Molla et al. Read next because GraphBit: A Graph-based Agentic Framework for Non-Linear Agent Orchestration overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partne…
[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…