TSR · The desk

Science · 2026

Articles in this category in 2026.

• TSR Desk · science · 26 October 2026, 07:00 UTC

  SWE-Shift: Evaluating Coding Agents in Evolving Workspaces

• TSR Desk · science · 26 October 2026, 07:00 UTC

  Reproducible Evaluation as the Substrate for Safe Prompt Compression in Production Agents

• TSR Desk · science · 26 October 2026, 07:00 UTC

  Async-RBench: Benchmarking Asynchronous Task Execution in LLM Agents

• TSR Desk · science · 26 October 2026, 07:00 UTC

  Minimum Description Length Unsupervisedly Separates Law-Grade from Statistical-Grade Knowledge

• TSR Desk · science · 26 October 2026, 07:00 UTC

  SWE-Clar: Can Coding Agents Clarify Incomplete Client Requirements and Deliver Greenfield

• TSR Desk · science · 26 October 2026, 07:00 UTC

  How Ground Truth Shapes Retrieval Scores: Evidence-Aware Evaluation of a Retrieval-Augmented

• TSR Desk · science · 26 October 2026, 07:00 UTC

  Guided Zero-Shot Anomaly Explanation for Anonymized Crowd Video Monitoring: Context,

• TSR Desk · science · 26 October 2026, 07:00 UTC

  Dementia-R1: Reinforced Pretraining and Reasoning from Unstructured Clinical Notes for

• TSR Desk · science · 26 October 2026, 07:00 UTC

  CEMIR vs. FISER: a comparative study of reasoning-based and enrichment-based approaches to

• TSR Desk · science · 26 October 2026, 07:00 UTC

  Uncertainty-Guided Latent Diagnostic Trajectory Learning for Sequential Clinical Diagnosis

• TSR Desk · science · 26 October 2026, 07:00 UTC

  BARC-Bench: Benign Asks, Risky Consequences

• TSR Desk · science · 26 October 2026, 07:00 UTC

  CASE: Confusion-Aware Skill Evolution for Frozen Multimodal Emotion Models

• TSR Desk · science · 26 October 2026, 07:00 UTC

  Tampered ML: Evaluating LLM Judgment of Machine Learning Research

• TSR Desk · science · 26 October 2026, 07:00 UTC

  Enhancing Table Recognition with Vision LLMs: A Benchmark and Neighbor-Guided Toolchain Reasoner

• TSR Desk · science · 26 October 2026, 07:00 UTC

  RubricCraft: Where Solutions Diverge, Criteria Emerge

• TSR Desk · science · 26 October 2026, 07:00 UTC

  MirrorCraft: Paired Evaluation under Hidden Rule Changes in Minecraft

• TSR Desk · science · 24 October 2026, 01:00 UTC

  WhisperRec: Latent Reasoning for Efficient Foundation Recommendation Models

• TSR Desk · science · 24 October 2026, 01:00 UTC

  FLAWS: A Benchmark for Error Identification and Localization in Scientific Papers

• TSR Desk · science · 24 October 2026, 01:00 UTC

  When LLMs Forget What They Believe: Characterizing State Integrity Collapse in Long-Horizon

• TSR Desk · science · 24 October 2026, 01:00 UTC

  TORUS: A Unified Closed-Loop Benchmark for Unified Audio Models

• TSR Desk · science · 24 October 2026, 01:00 UTC

  Self-Assessment Reinforcement Learning for Search-State Awareness in Retrieval-Augmented Large

• TSR Desk · science · 24 October 2026, 01:00 UTC

  ColaBench: Evaluating Memory-Guided Knowledge Acquisition in Personal Agents

• TSR Desk · science · 24 October 2026, 01:00 UTC

  From Clinical Judgment to Checkable Rules: Clinician-Led Guardrails for Mental-Health Chatbots

• TSR Desk · science · 24 October 2026, 01:00 UTC

  SCOPE: Compiling Natural-Language Policies into Neurosymbolic Predicate Programs

• TSR Desk · science · 24 October 2026, 01:00 UTC

  MAVOS-DD: Multilingual Audio-Video Open-Set Deepfake Detection Benchmark

• TSR Desk · science · 24 October 2026, 01:00 UTC

  TerraTransfer++: Demonstration-Free End-to-End Planner

• TSR Desk · science · 24 October 2026, 01:00 UTC

  Algorithmic drift in generative adaptive learning

• TSR Desk · science · 24 October 2026, 01:00 UTC

  EpiSEM-Bench: Evaluating Safety and Evidence Recovery in LLM Agents under Missing Evidence

• TSR Desk · science · 24 October 2026, 01:00 UTC

  Measuring the Permission Gate: A Stress-Test Evaluation of Claude Code's Auto Mode

• TSR Desk · science · 24 October 2026, 01:00 UTC

  What Lies Beyond Perception: A Partition Theory of the Limits of AI Evaluation and Oversight

• TSR Desk · science · 23 October 2026, 01:00 UTC

  CountVideoBench: A Benchmark to Evaluate Prediction Biases on the Object Counting task by

• TSR Desk · science · 23 October 2026, 01:00 UTC

  CoSec: Benchmarking Agent Security in Communities

• TSR Desk · science · 23 October 2026, 01:00 UTC

  Inheritance as an Admission Test for LLM Prompt Merge Operators

• TSR Desk · science · 23 October 2026, 01:00 UTC

  Closing the Sim-to-Real Gap: An Evaluation Framework for Autonomous Cyber Defense Configuration

• TSR Desk · science · 23 October 2026, 01:00 UTC

  HybridResearcher: Bridging the Structured-Unstructured Boundary for Deep Research Agents

• TSR Desk · science · 23 October 2026, 01:00 UTC

  SWE-DIVE: Multi-Turn Coding Agent Evaluation with Dynamic User Simulation

• TSR Desk · science · 23 October 2026, 01:00 UTC

  SANE: Automated Scorer-Specific Metamorphic Testing for LLM Judges

• TSR Desk · science · 23 October 2026, 01:00 UTC

  OwlSight: A Panoramic Monitor for Chain-of-Thought Reasoning

• TSR Desk · science · 23 October 2026, 01:00 UTC

  AnnoCaseLaw: Legal Annotation Benchmark of Civil Negligence

• TSR Desk · science · 23 October 2026, 01:00 UTC

  Environment Duels: A Microcosm of LLM Co-Evolution

• TSR Desk · science · 23 October 2026, 01:00 UTC

  What the Author Wrote, What the Agent Reads: Silent Cue Loss in Enterprise Document Ingestion

• TSR Desk · science · 23 October 2026, 01:00 UTC

  RegoMender: A Multi-agent Framework for Automated Rego Policy Repair

• TSR Desk · science · 23 October 2026, 01:00 UTC

  Your Benchmark Is an Empirical Measure Over Difficulty

• TSR Desk · science · 22 October 2026, 01:00 UTC

  Code2Video-Compass: A Comprehensive Benchmark for General-Purpose Code-Based Video Generation

• TSR Desk · science · 22 October 2026, 01:00 UTC

  MOSS: A Production Framework for Monitoring and Diagnosing Semantic Failures in Search and

• TSR Desk · science · 22 October 2026, 01:00 UTC

  WORTH Benchmark: Evidence Discovery under Not-Fully-Cooperative Information -- A Two-Episode

• TSR Desk · science · 22 October 2026, 01:00 UTC

  ABench: Adversarily Benchmarking Large Language Models across Taxonomized Values

• TSR Desk · science · 22 October 2026, 01:00 UTC

  Learning Computational Structure for Sample-Efficient Prediction

• TSR Desk · science · 22 October 2026, 01:00 UTC

  Navalia: Advancing Cowork Agents through Verifiable Task Synthesis and Long-Horizon

• TSR Desk · science · 22 October 2026, 01:00 UTC

  GenoMAS: Guided Multi-Agent Programming for Scientific Discovery from Genomic Data

• TSR Desk · science · 22 October 2026, 01:00 UTC

  Interruption-Aware Recurrent Reinforcement Learning for UAS Navigation Under Temporary Overrides

• TSR Desk · science · 22 October 2026, 01:00 UTC

  Give the Model a Finger: Developmentally Aligned Evaluation of Counting in Vision-Language

• TSR Desk · science · 22 October 2026, 01:00 UTC

  The Reward Model Selection Crisis in Personalized Alignment

• TSR Desk · science · 22 October 2026, 01:00 UTC

  Non-Stationarity Breaks Permutation Surrogates in Multi-Agent Reinforcement Learning: Diagnosis

• TSR Desk · science · 22 October 2026, 01:00 UTC

  CUA-Recovery: Benchmarking Error Recovery in Long-Horizon Computer-Use Agents

• TSR Desk · science · 22 October 2026, 01:00 UTC

  Repair the Core, Not the Symptom: Planning-Graph Repair for Long-Horizon Agents

• TSR Desk · science · 22 October 2026, 01:00 UTC

  Polarity Blindness in LLM Safety: A Systematic Study of Negation, Antonym Substitution, and

• TSR Desk · science · 22 October 2026, 01:00 UTC

  PReMISE: Policy Rubrics as Measurement Specifications for LLM Judges

• TSR Desk · science · 22 October 2026, 01:00 UTC

  SpectralTrain: A Universal Framework for Hyperspectral Image Classification

• TSR Desk · science · 22 October 2026, 01:00 UTC

  BAIT: Can LLMs Find What Biases a Reader? A Benchmark for Cognitive Bias Triggers in News

• TSR Desk · science · 22 October 2026, 01:00 UTC

  How Intervention History Changes AI Agent Sabotage on Subsequent Tasks

• TSR Desk · science · 21 October 2026, 01:00 UTC

  When Retrieval Partially Conflicts with Skills: Understanding Procedural Hallucination in

• TSR Desk · science · 21 October 2026, 01:00 UTC

  MoveBench: A Rubric-Based Benchmark for Evaluating Large Language Models in Sports and Health

• TSR Desk · science · 21 October 2026, 01:00 UTC

  Graph-Regularized Synthetic Control for Stable Counterfactual Estimation under Donor Redundancy

• TSR Desk · science · 21 October 2026, 01:00 UTC

  Benchmarking the Limits of In-Context Reinforcement Learning for Ad-Hoc Teamwork

• TSR Desk · science · 21 October 2026, 01:00 UTC

  CorrectRTL: Boost RTL Generation Quality Through Automated Self-Correction with Instruction

• TSR Desk · science · 21 October 2026, 01:00 UTC

  AI estimation of partition functions with the PULSE method for the thermodynamic properties of

• TSR Desk · science · 21 October 2026, 01:00 UTC

  VReAct: Verification as an Architectural Layer for LLM Agents: A V-Model Design, and a Pilot

• TSR Desk · science · 21 October 2026, 01:00 UTC

  Can We Trust Item Response Theory for AI Evaluation?

• TSR Desk · science · 21 October 2026, 01:00 UTC

  When the Editing Intent Is Split: A Cross-Modal Jailbreak Attack for Large Image Editing Models

• TSR Desk · science · 21 October 2026, 01:00 UTC

  Visual Text Compression as Measure Transport

• TSR Desk · science · 21 October 2026, 01:00 UTC

  IWOO: Wave Optics Optimizer Improved with Hybrid Strategies for Numerical Optimization and

• TSR Desk · science · 21 October 2026, 01:00 UTC · sooner

  VibeTraining-Bench: Data-Trainer-Harness Co-Design for Iterative System Improvement

• TSR Desk · science · 21 October 2026, 01:00 UTC

  Severity-Channel Conditioning: An Injection-Level Study for Consumer-Device Deblurring

• TSR Desk · science · 21 October 2026, 01:00 UTC

  Medical Fine-Tuning Reshapes, Not Reduces, Patient-Facing AI Risk: A Rubric-Based, LLM-Judge

• TSR Desk · science · 21 October 2026, 01:00 UTC

  InfraBench: Evaluating Infrastructure Agents Across Layers, Lifecycle, and Risk

• TSR Desk · science · 21 October 2026, 01:00 UTC

  Aligning Cultural Behaviors in Arabic Language Models via Preferences

• TSR Desk · science · 20 October 2026, 07:00 UTC

  Game of Phones: Harnessing Game Theory and LLMs in 'Bot Wars' to Counteract Phone Scams

• TSR Desk · science · 20 October 2026, 07:00 UTC

  Toollery: Scaling LLM Agents to Thousands of Skills and Tools

• TSR Desk · science · 20 October 2026, 07:00 UTC

  Pattern-Driven Evaluation of LLM Code Optimization: Isolating Compiler-Transcendent Semantic

• TSR Desk · science · 20 October 2026, 07:00 UTC

  VLA-Rolling: Fewer Denoising Steps, Better Control via Adaptive Action Reuse

• TSR Desk · science · 20 October 2026, 07:00 UTC

  Toward Collective Model Intelligence: Characterizing and Coordinating Models

• TSR Desk · science · 20 October 2026, 07:00 UTC

  Honeyed Words, Hidden Daggers: Benchmarking and Training LLMs to See Through Mock Politeness

• TSR Desk · science · 20 October 2026, 07:00 UTC

  AnyGoal: On World Model Generalization Beyond Expert Rollouts

• TSR Desk · science · 20 October 2026, 07:00 UTC

  Can LLM Agents Resolve Service Failures? Simulating Multi-Turn Complaint Handling under

• TSR Desk · science · 20 October 2026, 07:00 UTC

  KAYROS: An Anytime and Exact Open-Source Solver for Duration-Minimization Time-Dependent

• TSR Desk · science · 20 October 2026, 07:00 UTC

  CoworkEnv: Synthesizing Training Environments for Long-Horizon Cowork Agents

• TSR Desk · science · 20 October 2026, 07:00 UTC

  ConCI: A Controlled Counterfactual Benchmark for Contextual Integrity in LLMs

• TSR Desk · science · 20 October 2026, 07:00 UTC

  PDB: Not All Drivers Are the Same -- A Personalized Dataset for Understanding Driving Behavior

• TSR Desk · science · 20 October 2026, 07:00 UTC

  The Window Is in the Reasoning: When LLM Agents Decide to Follow Prompt Injections

• TSR Desk · science · 20 October 2026, 07:00 UTC

  Can LLM Agents Respond to Disasters? Benchmarking Heterogeneous Geospatial Reasoning in

• TSR Desk · science · 18 October 2026, 01:00 UTC

  AlphaDiana: A Measurement System for Agents in Verifiable Reasoning Tasks

• TSR Desk · science · 18 October 2026, 01:00 UTC

  ParanoiaEval: Benchmarking Unnecessary Defensive Work in Agentic Coding

• TSR Desk · science · 18 October 2026, 01:00 UTC

  CompMat-Bench: Benchmarking AI Agents for Computational Materials Science

• TSR Desk · science · 18 October 2026, 01:00 UTC

  DialFork: Going Beyond Factuality to Evaluate Procedural Hallucination in Conversational Agents

• TSR Desk · science · 18 October 2026, 01:00 UTC

  Frame Evidence:A Benchmark for Reference Frames and Visual Evidence in Multi-Image Spatial

• TSR Desk · science · 18 October 2026, 01:00 UTC

  In the Guise of Correction: How Framing and Cultural Adaptation Erode LLM Safety on Real-World

• TSR Desk · science · 18 October 2026, 01:00 UTC

  LIBERO-Industry: Benchmarking Robotic Manipulation under Industrial Process Constraints

• TSR Desk · science · 18 October 2026, 01:00 UTC

  SWE-Nexus: Automated Composition of Software Engineering Tasks from Test Execution

• TSR Desk · science · 18 October 2026, 01:00 UTC

  Filling Before Advancing: Capability-Gap-Driven Post-Training for Scenario-Specialized Remote