3/31/2026
Prototype-Based Semantic Consistency Alignment for Domain Adaptive Retrieval
Domain adaptive retrieval method using prototype-based semantic consistency alignment for transfer learning.
3/31/2026
A Semi Centralized Training Decentralized Execution Architecture for Multi Agent Deep Reinforcement Learning in Traffic Signal Control
Semi-centralized MARL architecture for traffic signal control combining centralized training and decentralized execution.
3/31/2026
Multilingual Medical Reasoning for Question Answering with Large Language Models
Method for multilingual medical QA using LLMs with reasoning traces from Wikipedia.
3/31/2026
Disrupting Hierarchical Reasoning: Adversarial Protection for Geographic Privacy in Multimodal Reasoning Models
ReasonBreak: Adversarial defense against privacy attacks in multimodal LLMs via geographic inference.
3/31/2026
BabyVLM-V2: Toward Developmentally Grounded Pretraining and Benchmarking of Vision Foundation Models
BabyVLM-V2 improves vision foundation model pretraining using developmental psychology principles and introduces DevCV Toolbox for cognitive evaluation.
3/31/2026
Does Tone Change the Answer? Evaluating Prompt Politeness Effects on Modern LLMs: GPT, Gemini, and LLaMA
Systematic evaluation of prompt politeness effects on GPT-4o mini, Gemini, and LLaMA shows linguistic tone influences LLM accuracy across model families.
3/31/2026
RadImageNet-VQA: A Large-Scale CT and MRI Dataset for Radiologic Visual Question Answering
RadImageNet-VQA is a large-scale dataset of 750K CT/MRI images with 7.5M QA pairs for advancing radiologic visual question answering.
3/31/2026
Measuring all the noises of LLM Evals
Measuring all the noises of LLM Evals defines and quantifies three types of noise in LLM evaluation to improve experimental signal-to-noise ratio.
3/31/2026
A Modular Reference Architecture for MCP-Servers Enabling Agentic BIM Interaction
Reference architecture for MCP-Servers enables LLM agents to interact with Building Information Models through standardized tool-calling interface.
3/31/2026
JMedEthicBench: A Multi-Turn Conversational Benchmark for Evaluating Medical Safety in Japanese Large Language Models
JMedEthicBench is a multi-turn conversational benchmark for evaluating medical safety in Japanese LLMs, addressing gaps in non-English medical evaluation.
3/31/2026
Vision-Language Agents for Interactive Forest Change Analysis
Vision-Language Agents for Interactive Forest Change Analysis integrates LLMs with vision-language models for satellite imagery analysis and semantic change captioning.
3/31/2026
Sparse-RL: Breaking the Memory Wall in LLM Reinforcement Learning via Stable Sparse Rollouts
Sparse-RL addresses memory bottlenecks in LLM reinforcement learning by proposing stable sparse KV cache rollouts for efficient training on limited hardware.
3/31/2026
An Agentic Operationalization of DISARM for FIMI Investigation on Social Media
Agentic approach to DISARM framework for investigating foreign information manipulation across social media platforms.
3/31/2026
LLMs versus the Halting Problem: Revisiting Program Termination Prediction
Analysis of LLM capabilities for predicting program termination, evaluating approximations to the undecidable Halting Problem.
3/31/2026
On the Impact of AGENTS.md Files on the Efficiency of AI Coding Agents
Empirical study on impact of AGENTS.md repository configuration files on runtime and token efficiency of autonomous AI coding agents.
3/31/2026
TextBFGS: A Case-Based Reasoning Approach to Code Optimization via Error-Operator Retrieval
TextBFGS case-based reasoning framework for iterative LLM code optimization leveraging past problem-solving experiences.
3/31/2026
Can Small Language Models Handle Context-Summarized Multi-Turn Customer-Service QA? A Synthetic Data-Driven Comparative Evaluation
Evaluation of small language models on multi-turn customer service QA using synthetic data compared to LLMs.
3/31/2026
SparVAR: Exploring Sparsity in Visual AutoRegressive Modeling for Training-Free Acceleration
Training-free acceleration method for Visual AutoRegressive modeling via sparsity exploration to reduce attention complexity.
3/31/2026
Unveiling Implicit Advantage Symmetry: Why GRPO Struggles with Exploration and Difficulty Adaptation
Theoretical analysis of GRPO reinforcement learning for LLM reasoning, identifying implicit advantage symmetry limitations in exploration and difficulty adaptation.
3/31/2026
A Theoretical Analysis of Test-Driven LLM Code Generation
Theoretical probabilistic framework analyzing test-driven LLM code generation strategies for environment interaction and selection heuristics.
3/31/2026
CLEAR: A Knowledge-Centric Vessel Trajectory Analysis Platform
Knowledge-centric vessel trajectory analysis platform leveraging LLMs to transform raw maritime AIS data for expert analysis.
3/31/2026
CoPE-VideoLM: Leveraging Codec Primitives For Efficient Video Language Modeling
Video language model using codec primitives for efficient temporal understanding within context window constraints.
3/31/2026
When AI Agents Teach Each Other: Discourse Patterns Resembling Peer Learning in the Moltbook Community
Analysis of 2.4 million AI agents in Moltbook community exhibiting peer learning discourse patterns and collaborative knowledge sharing.
3/31/2026
MALLVI: A Multi-Agent Framework for Integrated Generalized Robotics Manipulation
Multi-agent LLM and vision framework for closed-loop robotic manipulation with environmental feedback for task planning.
3/31/2026
FUSAR-GPT : A Spatiotemporal Feature-Embedded and Two-Stage Decoupled Visual Language Model for SAR Imagery
Visual language model adapted for SAR imagery interpretation using spatiotemporal feature embedding and two-stage decoupled architecture.
3/31/2026
PhysMem: Scaling Test-time Physical Memory for Robot Manipulation
PhysMem framework enables vision-language model robot planners to leverage physical memory for improved object manipulation and reasoning about material properties.
3/31/2026
Efficient Encoder-Free Fourier-based 3D Large Multimodal Model
3D large multimodal model using Fourier-based tokenization to process point clouds without heavy pre-trained encoders for improved efficiency.
3/31/2026
AG-VAS: Anchor-Guided Zero-Shot Visual Anomaly Segmentation with Large Multimodal Models
AG-VAS uses large multimodal models for zero-shot visual anomaly segmentation with anchor-guided approach to align semantic and spatial features.
3/31/2026
MetaState: Persistent Working Memory Enhances Reasoning in Discrete Diffusion Language Models
MetaState enhances discrete diffusion language models by preserving intermediate continuous representations across denoising steps to improve reasoning performance.
3/31/2026
Towards Privacy-Preserving LLM Inference via Covariant Obfuscation (Technical Report)
Privacy-preserving LLM inference technique using covariant obfuscation to protect data during cloud-based inference while maintaining accuracy and efficiency.
3/31/2026
Thin Keys, Full Values: Reducing KV Cache via Low-Dimensional Attention Selection
Research on reducing KV cache in Transformer attention by using low-dimensional keys for token selection while maintaining full-dimensional values, achieving O(log N) efficiency.
3/31/2026
Image Generation Models: A Technical History
Comprehensive technical survey of image generation models covering VAEs, GANs, flows, autoregressive, transformers, and diffusion methods.
3/31/2026
Distributional Regression with Tabular Foundation Models: Evaluating Probabilistic Predictions via Proper Scoring Rules
Evaluation framework for tabular foundation models using proper scoring rules to assess predicted distributions beyond point estimates.
3/31/2026
Understanding the Use of a Large Language Model-Powered Guide to Make Virtual Reality Accessible for Blind and Low Vision People
User study of LLM-powered sighted guide for blind and low vision users navigating social virtual reality environments.
3/31/2026
Feedback-Coupled Memory Systems: A Dynamical Model for Adaptive Coordination
Dynamical framework for adaptive coordination in multi-agent systems using feedback-coupled memory systems.
3/31/2026
AgentDrift: Unsafe Recommendation Drift Under Tool Corruption Hidden by Ranking Metrics in LLM Agents
AgentDrift identifies safety vulnerabilities in tool-augmented LLM agents through paired-trajectory protocol testing under tool corruption.
3/31/2026
Real-World AI Evaluation: How FRAME Generates Systematic Evidence to Resolve the Decision-Maker's Dilemma
FRAME methodology for real-world AI evaluation generating systematic evidence of system behavior across diverse deployment contexts.
3/31/2026
GhanaNLP Parallel Corpora: Comprehensive Multilingual Resources for Low-Resource Ghanaian Languages
GhanaNLP parallel corpora dataset with 41,513 sentence pairs for five low-resource Ghanaian languages: Twi, Fante, Ewe, Ga, Kusaal.
3/31/2026
Deconfounded Lifelong Learning for Autonomous Driving via Dynamic Knowledge Spaces
DeLL framework for lifelong learning in autonomous driving using Dirichlet process mixture models to address catastrophic forgetting.
3/31/2026
EngGPT2: Sovereign, Efficient and Open Intelligence
EngGPT2-16B Italian LLM achieving competitive performance on MMLU-Pro, GSM8K, and HumanEval with 5-50% lower inference cost.
3/31/2026
InCoder-32B: Code Foundation Model for Industrial Scenarios
InCoder-32B code foundation model optimized for industrial programming tasks with hardware semantics and resource constraints.
3/31/2026
Scaling Sim-to-Real Reinforcement Learning for Robot VLAs with Generative 3D Worlds
Sim-to-real reinforcement learning approach for vision-language-action robot models using generative 3D world environments.
3/31/2026
HypeLoRA: Hyper-Network-Generated LoRA Adapters for Calibrated Language Model Fine-Tuning
HypeLoRA framework using hyper-networks for parameter-efficient fine-tuning of language models with improved calibration on GLUE.
3/31/2026
Modernizing Amdahl's Law: How AI Scaling Laws Shape Computer Architecture
Reformulation of Amdahl's Law for modern heterogeneous systems with AI scaling dynamics and resource constraints.
3/31/2026
Alignment Whack-a-Mole : Finetuning Activates Verbatim Recall of Copyrighted Books in Large Language Models
Study showing finetuning bypasses alignment safeguards causing LLMs to verbatim recall copyrighted training data.
3/31/2026
LLM-Powered Workflow Optimization for Multidisciplinary Software Development: An Automotive Industry Case Study
LLM-powered workflow optimization for multidisciplinary software development in automotive industry bridging domain experts and developers.
3/31/2026
KG-Hopper: Empowering Compact Open LLMs with Knowledge Graph Reasoning via Reinforcement Learning
KG-Hopper: Framework enabling compact open LLMs to perform multi-hop knowledge graph reasoning via reinforcement learning.
3/31/2026
Scaling Attention via Feature Sparsity
Sparse Feature Attention: Method to reduce transformer self-attention complexity via feature sparsity instead of sequence-level sparsity.
3/31/2026
Code Review Agent Benchmark
Code Review Agent Benchmark: Dataset for evaluating AI agents on code quality assurance and review tasks.
3/31/2026
Synthetic Mixed Training: Scaling Parametric Knowledge Acquisition Beyond RAG
Synthetic Mixed Training method combining synthetic QA and document generation to improve LLM knowledge acquisition beyond RAG performance.