2/17/2026
MURPHY: Multi-Turn GRPO for Self Correcting Code Generation
MURPHY: Multi-turn reinforcement learning framework for self-correcting code generation combining group relative policy optimization with execution verification.
2/17/2026
Evolution Strategies at the Hyperscale
EGGROLL: Scalable evolution strategies algorithm using low-rank approximations to improve training efficiency of black-box optimization on GPUs.
2/17/2026
ESPO: Entropy Importance Sampling Policy Optimization
ESPO: Entropy importance sampling policy optimization for stable and efficient token-level RL training of LLMs on complex reasoning tasks at scale.
2/17/2026
RRPO: Robust Reward Policy Optimization for LLM-based Emotional TTS
RRPO: Robust reward policy optimization framework preventing reward hacking in LLM-based emotional text-to-speech by addressing vulnerability of vanilla reward models.
2/17/2026
ArtistMus: A Globally Diverse, Artist-Centric Benchmark for Retrieval-Augmented Music Question Answering
ArtistMus: Benchmark dataset for retrieval-augmented music question answering grounded in artist metadata to evaluate LLMs on music-related reasoning tasks.
2/17/2026
IntentMiner: Intent Inversion Attack via Tool Call Analysis in the Model Context Protocol
IntentMiner: Privacy attack exploiting Model Context Protocol servers to extract user intents from LLM tool calls, revealing new security vulnerabilities in agentic AI systems.
2/17/2026
Emergent human-like working memory from artificial neurons with intrinsic plasticity
IPNet: Neuromorphic architecture using magnetic tunnel junction intrinsic plasticity to implement human-like working memory with reduced energy costs.
2/17/2026
Adaptive Agents in Spatial Double-Auction Markets: Modeling the Emergence of Industrial Symbiosis
Agent-based model simulating adaptive firm behavior in spatial double-auction markets to understand emergence of industrial symbiosis under socio-spatial constraints.
2/17/2026
Multi-LLM Thematic Analysis with Dual Reliability Metrics: Combining Cohen's Kappa and Semantic Similarity for Qualitative Research Validation
Multi-LLM validation framework for thematic analysis combining Cohen's Kappa and semantic similarity metrics to improve reliability of LLM-based qualitative research coding.
2/17/2026
Nightjar: Dynamic Adaptive Speculative Decoding for Large Language Models Serving
Nightjar: Dynamic adaptive speculative decoding method that adjusts verification overhead based on request load to optimize LLM inference throughput and latency.
2/17/2026
Context Volume Drives Performance: Tackling Domain Shift in Extremely Low-Resource Translation via RAG
Retrieval-augmented generation approach addressing domain shift in low-resource neural machine translation using context volume from limited corpora.
2/17/2026
Orthogonalized Policy Optimization:Decoupling Sampling Geometry from Optimization Geometry in RLHF
Unified framework for LLM alignment decoupling sampling and optimization geometry across PPO, DPO, IPO algorithms and variants.
2/17/2026
Improved Bug Localization with AI Agents Leveraging Hypothesis and Dynamic Cognition
LLM-based AI agents with hypothesis-driven cognition for improved software bug localization by analyzing code component relationships.
2/17/2026
An Agentic Operationalization of DISARM for FIMI Investigation on Social Media
Agentic operationalization of DISARM framework for investigating foreign information manipulation on social media across NATO allied partners.
2/17/2026
GRIP: Algorithm-Agnostic Machine Unlearning for Mixture-of-Experts via Geometric Router Constraints
Machine unlearning method for Mixture-of-Experts LLMs using geometric router constraints to erase knowledge rather than redirect queries.
2/17/2026
Interpreting Emergent Extreme Events in Multi-Agent Systems
Framework for interpreting and explaining emergent extreme events in LLM-powered multi-agent systems to improve safety and transparency.
2/17/2026
Reinforcement Learning via Self-Distillation
Self-distillation approach for reinforcement learning leveraging rich textual feedback from verifiable environments to improve credit assignment in code/math tasks.
2/17/2026
ShotFinder: Imagination-Driven Open-Domain Video Shot Retrieval via Web Search
Benchmark for open-domain video shot retrieval using LLMs for understanding editing requirements and retrieving keyframe-oriented shots.
2/17/2026
From Associations to Activations: Comparing Behavioral and Hidden-State Semantic Geometry in LLMs
Analyzing semantic geometry in LLM hidden states versus behavioral similarity through psycholinguistic experiments across eight instruction-tuned models.
2/17/2026
Unifying Ranking and Generation in Query Auto-Completion via Retrieval-Augmented Generation and Multi-Objective Alignment
Unified retrieval-augmented generation framework for query auto-completion combining ranking and generation to reduce hallucination and improve coverage.
2/17/2026
Reinforcement Learning with Promising Tokens for Large Language Models
Reinforcement learning technique filtering irrelevant tokens to improve LLM policy optimization by focusing on contextually relevant action spaces.
2/17/2026
Accelerating Scientific Research with Gemini: Case Studies and Common Techniques
Case studies of Google's Gemini models assisting scientific research including mathematical discovery and routine task automation.
2/17/2026
Privileged Information Distillation for Language Models
Studying knowledge distillation from privileged information in language models for multi-turn agentic environments, addressing inference-time capability transfer.
2/17/2026
On the Non-Identifiability of Steering Vectors in Large Language Models
Demonstrates that steering vectors in LLMs are fundamentally non-identifiable due to large equivalence classes of behaviorally identical vectors.
2/17/2026
ShallowJail: Steering Jailbreaks against Large Language Models
Steering-based jailbreak method against aligned LLMs requiring less computation than white-box approaches but maintaining stealth.
2/17/2026
Fin-RATE: A Real-world Financial Analytics and Tracking Evaluation Benchmark for LLMs on SEC Filings
Benchmark for evaluating LLM performance on financial analysis and tracking using SEC filings with multi-document synthesis.
2/17/2026
Debugging code world models
Analyzes errors and limitations in Code World Models that simulate program execution by predicting runtime state.
2/17/2026
Language Modeling and Understanding Through Paraphrase Generation and Detection
Studies language understanding through paraphrase generation and detection capabilities in language models.
2/17/2026
Predictive Query Language: A Domain-Specific Language for Predictive Modeling on Relational Databases
Domain-specific language for predictive modeling on relational databases covering missing values and future predictions.
2/17/2026
Kunlun: Establishing Scaling Laws for Massive-Scale Recommendation Systems through Unified Architecture Design
Derives scaling laws for massive-scale recommendation systems through unified architecture design and efficiency improvements.
2/17/2026
Anonymization-Enhanced Privacy Protection for Mobile GUI Agents: Available but Invisible
Privacy protection method for mobile GUI agents using anonymization to mitigate exposure of sensitive data during screen processing.
2/17/2026
C^2ROPE: Causal Continuous Rotary Positional Encoding for 3D Large Multimodal-Models Reasoning
Causal continuous rotary positional encoding improving 3D feature alignment in multimodal LLMs for spatial reasoning tasks.
2/17/2026
LAP: Language-Action Pre-Training Enables Zero-shot Cross-Embodiment Transfer
Language-Action pre-training framework enabling zero-shot transfer of robot policies across different embodiments without fine-tuning.
2/17/2026
In-the-Wild Model Organisms: Mitigating Undesirable Emergent Behaviors in Production LLM Post-Training via Data Attribution
Data attribution method to trace and mitigate undesirable emergent behaviors in LLM post-training by identifying responsible datapoints.
2/17/2026
ReaDy-Go: Real-to-Sim Dynamic 3D Gaussian Splatting Simulation for Environment-Specific Visual Navigation with Moving Obstacles
3D Gaussian Splatting simulation framework for training visual navigation policies in dynamic environments with moving obstacles.
2/17/2026
Zooming without Zooming: Region-to-Image Distillation for Fine-Grained Multimodal Perception
Method to improve fine-grained visual perception in multimodal LLMs by distilling zoomed regions without repeated inference calls.
2/17/2026
ExtractBench: A Benchmark and Evaluation Methodology for Complex Structured Extraction
Benchmark and methodology for evaluating LLM-based PDF-to-JSON extraction at enterprise scale with complex schema requirements.
2/17/2026
Agent Skills for Large Language Models: Architecture, Acquisition, Security, and the Path Forward
Formalizes agent skills as composable packages enabling LLMs to extend capabilities dynamically without retraining, covering architecture, acquisition, and security.
2/17/2026
MedXIAOHE: A Comprehensive Recipe for Building Medical MLLMs
MedXIAOHE medical vision-language foundation model with entity-aware continual pretraining achieving state-of-the-art on medical benchmarks.
2/17/2026
Prior-Guided Symbolic Regression: Towards Scientific Consistency in Equation Discovery
Prior-guided symbolic regression framework incorporating scientific constraints to discover interpretable equations consistent with physical principles.
2/17/2026
Directional Concentration Uncertainty: A representational approach to uncertainty quantification for generative models
Directional Concentration Uncertainty framework for flexible uncertainty quantification in generative models across tasks and modalities.
2/17/2026
BLUEPRINT Rebuilding a Legacy: Multimodal Retrieval for Complex Engineering Drawings and Documents
Blueprint system for multimodal retrieval in engineering archives using layout-aware VLM OCR and identifier normalization.
2/17/2026
Exploring the Performance of ML/DL Architectures on the MNIST-1D Dataset
Comparative study of ML/DL architectures on MNIST-1D dataset for distinguishing neural network performance on controlled benchmarks.
2/17/2026
The Speed-up Factor: A Quantitative Multi-Iteration Active Learning Performance Metric
Review and proposed metrics for evaluating multi-iteration active learning query methods and sample selection strategies.
2/17/2026
Finding Highly Interpretable Prompt-Specific Circuits in Language Models
Research showing language model circuits are prompt-specific within tasks, not uniform across prompts, using causal communication analysis.
2/17/2026
Federated Learning of Nonlinear Temporal Dynamics with Graph Attention-based Cross-Client Interpretability
Federated learning framework for nonlinear temporal dynamics using graph attention for interpretable cross-client relationships without sharing raw data.
2/17/2026
Preventing Rank Collapse in Federated Low-Rank Adaptation with Client Heterogeneity
Research identifies rank collapse phenomenon in federated low-rank adaptation with heterogeneous clients, proposing mitigation for privacy-preserving fine-tuning.
2/17/2026
TrasMuon: Trust-Region Adaptive Scaling for Orthogonalized Momentum Optimizers
TrasMuon optimizer enhances Muon-style methods with trust-region adaptive scaling to address training sensitivity and high-energy bursts.
2/17/2026
$\gamma$-weakly $\theta$-up-concavity: Linearizable Non-Convex Optimization with Applications to DR-Submodular and OSS Functions
Novel optimization framework for monotone non-convex functions unifying DR-submodular and OSS functions with theoretical convergence analysis.
2/17/2026
Singular Vectors of Attention Heads Align with Features
Study demonstrating singular vectors of attention heads align with learned features in language models, providing mechanistic interpretability foundations.