4/2/2026
Graceful Forgetting in Generative Language Models
Proposes graceful forgetting methods to mitigate negative transfer by selectively forgetting detrimental pre-training knowledge during fine-tuning of language models.
4/2/2026
How Does Alignment Enhance LLMs' Multilingual Capabilities? A Language Neurons Perspective
Analyzes language-specific neurons to understand how multilingual alignment transfers capabilities from high-resource to low-resource languages in LLMs.
4/2/2026
Two-stage Vision Transformers and Hard Masking offer Robust Object Representations
Two-stage vision transformer with hard masking approach for robust object representations that balance context dependence with distribution shift robustness.
4/2/2026
"Is This Really a Human Peer Supporter?": Misalignments Between Peer Supporters and Experts in LLM-Supported Interactions
Investigates misalignments between LLM-supported peer supporters and mental health experts, examining quality and safety concerns in AI-driven psychosocial support.
4/2/2026
MemeMind: A Large-Scale Multimodal Dataset with Chain-of-Thought Reasoning for Harmful Meme Detection
MemeMind dataset with chain-of-thought reasoning for detecting harmful memes, addressing implicit harmful content in multimodal text-image combinations.
4/2/2026
Klear-Reasoner: Advancing Reasoning Capability via Gradient-Preserving Clipping Policy Optimization
Klear-Reasoner model demonstrates long reasoning capabilities with gradient-preserving clipping for policy optimization, achieving strong benchmark performance with reproducible training details.
4/2/2026
Polychromic Objectives for Reinforcement Learning
Addresses mode collapse in reinforcement learning fine-tuning by introducing polychromic objectives that preserve policy diversity and enable better exploration.
4/2/2026
Adaptive Data-Knowledge Alignment in Genetic Perturbation Prediction
Proposes end-to-end integration of data-driven learning and existing knowledge for predicting transcriptional responses to genetic perturbations in biological systems.
4/2/2026
Are Large Vision-Language Models Ready to Guide Blind and Low-Vision Individuals?
Evaluates whether large vision-language models can effectively guide blind and low-vision individuals, addressing how to measure real-world utility beyond standard metrics.
4/2/2026
TempoControl: Temporal Attention Guidance for Text-to-Video Models
TempoControl method enables fine-grained temporal control in text-to-video generative models, allowing specification of when visual elements appear in sequences without retraining.
4/2/2026
Incoherence in Goal-Conditioned Autoregressive Models
Mathematical analysis of incoherence in goal-conditioned autoregressive models fine-tuned with reinforcement learning.
4/2/2026
BIOGEN: Evidence-Grounded Multi-Agent Reasoning Framework for Transcriptomic Interpretation in Antimicrobial Resistance
Multi-agent reasoning framework for interpreting gene clusters in antimicrobial resistance studies using transcriptomic data.
4/2/2026
E-Scores for (In)Correctness Assessment of Generative Model Outputs
Conformal prediction framework for assessing correctness of LLM outputs with user-defined tolerance levels.
4/2/2026
Benchmarking Educational LLMs with Analytics: A Case Study on Gender Bias in Feedback
Benchmarking framework using embeddings to detect gender bias in LLMs used for educational feedback on student essays.
4/2/2026
Structured Prompts Improve Evaluation of Language Models
Study showing structured prompts significantly improve LLM evaluation accuracy and reduce prompt-dependent variance in benchmark frameworks like HELM.
4/2/2026
OmniFusion: Simultaneous Multilingual Multimodal Translations via Modular Fusion
OmniFusion modular approach for simultaneous multilingual multimodal translation combining speech recognition and translation in open-source LLM pipelines.
4/2/2026
Lumos: Let there be Language Model System Certification
Lumos framework for formally certifying language model system behaviors using imperative probabilistic programming with graph-based prompt generation.
4/2/2026
Bypassing Prompt Injection Detectors through Evasive Injections
Study demonstrating evasive injection techniques that bypass ML-based prompt injection detectors in retrieval-augmented LLM systems.
4/2/2026
On the Non-Identifiability of Steering Vectors in Large Language Models
Analysis showing steering vectors in LLMs are fundamentally non-identifiable with large equivalence classes, questioning interpretability of activation steering methods.
4/2/2026
FIRE: Frobenius-Isometry Reinitialization for Balancing the Stability-Plasticity Tradeoff
FIRE reinitialization method balancing stability-plasticity tradeoff in continual learning for deep neural networks through Frobenius-isometry constraints.
4/2/2026
Evaluating LLM-Generated ACSL Annotations for Formal Verification
Empirical evaluation of LLM-generated ACSL formal specification annotations for C programs, assessing automatic verification without human assistance.
4/2/2026
Chat-Based Support Alone May Not Be Enough: Comparing Conversational and Embedded LLM Feedback for Mathematical Proof Learning
Empirical evaluation of GPTutor LLM tutoring system comparing embedded proof-review feedback versus chatbot support for discrete mathematics learning.
4/2/2026
SWE-CI: Evaluating Agent Capabilities in Maintaining Codebases via Continuous Integration
SWE-CI benchmark evaluating LLM-powered agents on repository-level codebase maintenance via continuous integration and multi-step feature iterations.
4/2/2026
RoboClaw: An Agentic Framework for Scalable Long-Horizon Robotic Tasks
RoboClaw agentic framework unifying data collection, policy learning, and deployment for long-horizon robotic tasks with vision-language-action systems.
4/2/2026
OPERA: Online Data Pruning for Efficient Retrieval Model Adaptation
OPERA framework for data pruning in dense retrieval models that improves both efficiency and effectiveness of domain-specific finetuning through heterogeneous pair selection.
4/2/2026
How Motivation Relates to Generative AI Use: A Large-Scale Survey of Mexican High School Students
Survey of 6,793 Mexican high school students examining how different motivational profiles relate to generative AI tool usage in math and writing.
4/2/2026
AI Agents Can Already Autonomously Perform Experimental High Energy Physics
Demonstrates LLM-based AI agents autonomously executing high energy physics analysis pipelines including event selection, background estimation, and statistical testing.
4/2/2026
Children's Intelligence Tests Pose Challenges for MLLMs? KidGym: A 2D Grid-Based Reasoning Benchmark for MLLMs
KidGym benchmark dataset based on children's intelligence tests to evaluate multimodal LLMs on visual reasoning tasks.
4/2/2026
Large Language Model Guided Incentive Aware Reward Design for Cooperative Multi-Agent Reinforcement Learning
Framework using LLMs to automate reward design for multi-agent reinforcement learning by synthesizing executable reward programs.
4/2/2026
Experiential Reflective Learning for Self-Improving LLM Agents
Experiential Reflective Learning framework enabling LLM agents to self-improve by leveraging past interactions and adapting to specialized environments.
4/2/2026
Closing the Confidence-Faithfulness Gap in Large Language Models
Mechanistic interpretability analysis of how LLMs verbalize confidence scores versus actual accuracy using linear probes and activation steering.
4/2/2026
Neuro-Symbolic Process Anomaly Detection
Neuro-symbolic approach combining neural networks with domain knowledge for process anomaly detection in event logs.
4/2/2026
Vision2Web: A Hierarchical Benchmark for Visual Website Development with Agent Verification
Vision2Web: Hierarchical benchmark for evaluating AI agents on website development tasks from UI-to-code to full-stack implementation.
4/2/2026
CarbonEdge: Carbon-Aware Deep Learning Inference Framework for Sustainable Edge Computing
CarbonEdge: Carbon-aware deep learning inference framework for edge computing optimizing environmental impact alongside latency and throughput.
4/2/2026
CDH-Bench: A Commonsense-Driven Hallucination Benchmark for Evaluating Visual Fidelity in Vision-Language Models
CDH-Bench: Benchmark evaluating vision-language models' commonsense-driven hallucinations when visual evidence conflicts with common sense.
4/2/2026
HISA: Efficient Hierarchical Indexing for Fine-Grained Sparse Attention
HISA improves efficiency of sparse attention mechanisms by optimizing hierarchical indexing to reduce bottlenecks in token-level key selection for LLMs.
4/2/2026
MemFactory: Unified Inference & Training Framework for Agent Memory
MemFactory: unified inference and training framework for agent memory integration with RL optimization of memory operations in LLM agents.
4/2/2026
Automatic Method Illustration Generation for AI Scientific Papers via Drawing Middleware Creation, Evolution, and Orchestration
FigAgent: multi-agent framework for automatic method illustration generation in AI papers via drawing middleware orchestration.
4/2/2026
From Density Matrices to Phase Transitions in Deep Learning: Spectral Early Warnings and Interpretability
Reduced density matrix method from quantum chemistry for predicting and interpreting phase transitions during deep learning model training.
4/2/2026
Two-Stage Optimizer-Aware Online Data Selection for Large Language Models
Optimizer-aware gradient-based online data selection framework for sequential LLM fine-tuning with step-dependent sample utility estimation.
4/2/2026
Task-Centric Personalized Federated Fine-Tuning of Language Models
Personalized federated fine-tuning approach for language models on distributed heterogeneous task datasets with improved generalization.
4/2/2026
Evolution Strategies for Deep RL pretraining
Evolution strategies for Deep RL pretraining offering derivative-free, computationally efficient alternative to standard deep reinforcement learning.
4/2/2026
Temporal Memory for Resource-Constrained Agents: Continual Learning via Stochastic Compress-Add-Smooth
Continual learning framework for resource-constrained agents using stochastic bridge diffusion process for temporal memory management.
4/2/2026
Perspective: Towards sustainable exploration of chemical spaces with machine learning
Perspective on sustainability challenges in AI-driven molecular and materials discovery across QM data, training, and automation pipelines.
4/2/2026
Empirical Validation of the Classification-Verification Dichotomy for AI Safety Gates
Empirical evidence that classifier-based safety gates fail for self-improving AI systems across multiple model architectures.
4/2/2026
Learning to Play Blackjack: A Curriculum Learning Perspective
LLM-based curriculum learning framework for reinforcement learning agents applied to Blackjack game strategy.
4/2/2026
Speeding Up Mixed-Integer Programming Solvers with Sparse Learning for Branching
Sparse interpretable machine learning models for improving branching decisions in mixed-integer programming solvers without GPU requirements.
4/2/2026
ParetoBandit: Budget-Paced Adaptive Routing for Non-Stationary LLM Serving
Open-source adaptive router for multi-model LLM serving using cost-aware contextual bandits with non-stationary pricing and quality changes.
4/2/2026
Offline Constrained RLHF with Multiple Preference Oracles
Offline reinforcement learning from human feedback with multiple preference oracles for trading off performance with safety and fairness constraints.
4/2/2026
Diversity-Aware Reverse Kullback-Leibler Divergence for Large Language Model Distillation
Diversity-aware reverse KL divergence method improving LLM distillation with large capacity mismatches.