AI Onderzoekspapers Dagelijks

Emerging Properties in Unified Multimodal Pretraining

SageAttention3: Microscaling FP4 Attention for Inference and An Exploration of 8-Bit Training

Optimizing Anytime Reasoning via Budget Relative Policy Optimization

VisualQuality-R1: Reasoning-Induced Image Quality Assessment via Reinforcement Learning to Rank

Neurosymbolic Diffusion Models

Reward Reasoning Model

Visual Agentic Reinforcement Fine-Tuning

Latent Flow Transformer

The Aloe Family Recipe for Open and Specialized Healthcare LLMs

Think Only When You Need with Large Hybrid-Reasoning Models

Reasoning Models Better Express Their Confidence

General-Reasoner: Advancing LLM Reasoning Across All Domains

Reasoning Path Compression: Compressing Generation Trajectories for Efficient LLM Reasoning

Visionary-R1: Mitigating Shortcuts in Visual Reasoning with Reinforcement Learning

Exploring Federated Pruning for Large Language Models

Training-Free Watermarking for Autoregressive Image Generation

VideoEval-Pro: Robust and Realistic Long Video Understanding Evaluation

Hunyuan-Game: Industrial-grade Intelligent Game Creation Model

Fine-tuning Quantized Neural Networks with Zeroth-order Optimization

CS-Sum: A Benchmark for Code-Switching Dialogue Summarization and the Limits of Large Language Models

SSR: Enhancing Depth Perception in Vision-Language Models via Rationale-Guided Spatial Reasoning

Two Experts Are All You Need for Steering Thinking: Reinforcing Cognitive Effort in MoE Reasoning Models Without Additional Training

NExT-Search: Rebuilding User Feedback Ecosystem for Generative AI Search

Visual Instruction Bottleneck Tuning

Lessons from Defending Gemini Against Indirect Prompt Injections

Not All Correct Answers Are Equal: Why Your Distillation Source Matters

Towards eliciting latent knowledge from LLMs with mechanistic interpretability

The Hallucination Tax of Reinforcement Finetuning

Vox-Profile: A Speech Foundation Model Benchmark for Characterizing Diverse Speaker and Speech Traits

Warm Up Before You Train: Unlocking General Reasoning in Resource-Constrained Settings

Fixing 7,400 Bugs for 1$: Cheap Crash-Site Program Repair

Truth Neurons

Phare: A Safety Probe for Large Language Models

MIGRATION-BENCH: Repository-Level Code Migration Benchmark from Java 8

CompeteSMoE -- Statistically Guaranteed Mixture of Experts Training via Competition

Solve-Detect-Verify: Inference-Time Scaling with Flexible Generative Verifier

Rethinking Optimal Verification Granularity for Compute-Efficient Test-Time Scaling

Bidirectional LMs are Better Knowledge Memorizers? A Benchmark for Real-world Knowledge Injection

Will AI Tell Lies to Save Sick Children? Litmus-Testing AI Values Prioritization with AIRiskDilemmas

CoIn: Counting the Invisible Reasoning Tokens in Commercial Opaque LLM APIs

To Bias or Not to Bias: Detecting bias in News with bias-detector

Incorporating brain-inspired mechanisms for multimodal learning in artificial intelligence

Tokenization Constraints in LLMs: A Study of Symbolic and Arithmetic Reasoning Limits

GeoRanker: Distance-Aware Ranking for Worldwide Image Geolocalization

Learning to Highlight Audio by Watching Movies

Masking in Multi-hop QA: An Analysis of How Language Models Perform with Context Permutation

KERL: Knowledge-Enhanced Personalized Recipe Recommendation using Large Language Models

Dynadiff: Single-stage Decoding of Images from Continuously Evolving fMRI

Understanding Gen Alpha Digital Language: Evaluation of LLM Safety Systems for Content Moderation

Void in Language Models

Towards Embodied Cognition in Robots via Spatially Grounded Synthetic Worlds

Object-Centric Representations Improve Policy Generalization in Robot Manipulation

The Distracting Effect: Understanding Irrelevant Passages in RAG