AI Research Papers Daily

Qwen2.5-VL Technical Report

On the Trustworthiness of Generative Foundation Models: Guideline, Assessment, and Perspective

SongGen: A Single Stage Auto-regressive Transformer for Text-to-Song Generation

RAD: Training an End-to-End Driving Policy via Large-Scale 3DGS-based Reinforcement Learning

MoM: Linear Sequence Modeling with Mixture-of-Memories

Craw4LLM: Efficient Web Crawling for LLM Pretraining

MMTEB: Massive Multilingual Text Embedding Benchmark

Small Models Struggle to Learn from Strong Reasoners

LongPO: Long Context Self-Evolution of Large Language Models through Short-to-Long Preference Optimization

Autellix: An Efficient Serving Engine for LLM Agents as General Programs

Thinking Preference Optimization

SearchRAG: Can Search Engines Be Helpful for LLM-based Medical Question Answering?

Presumed Cultural Identity: How Names Shape LLM Responses

Why Safeguarded Ships Run Aground? Aligned Large Language Models' Safety Mechanisms Tend to Be Anchored in The Template Region

AIDE: AI-Driven Exploration in the Space of Code

InfiR : Crafting Effective Small Language Models and Multimodal Small Language Models in Reasoning

NExT-Mol: 3D Diffusion Meets 1D Language Modeling for 3D Molecule Generation

AdaptiveStep: Automatically Dividing Reasoning Step through Model Confidence

Train Small, Infer Large: Memory-Efficient LoRA Training for Large Language Models

From Tools to Teammates: Evaluating LLMs in Multi-Session Coding Interactions

TESS 2: A Large-Scale Generalist Diffusion Language Model

REALTALK: A 21-Day Real-World Dataset for Long-Term Conversation

MVL-SIB: A Massively Multilingual Vision-Language Benchmark for Cross-Modal Topical Matching

REFIND: Retrieval-Augmented Factuality Hallucination Detection in Large Language Models

GIMMICK -- Globally Inclusive Multimodal Multitask Cultural Knowledge Benchmarking

ActionPiece: Contextually Tokenizing Action Sequences for Generative Recommendation

Which of These Best Describes Multiple Choice Evaluation with LLMs? A) Forced B) Flawed C) Fixable D) All of the Above

Judging the Judges: A Collection of LLM-Generated Relevance Judgements

Reducing Hallucinations in Language Model-based SPARQL Query Generation Using Post-Generation Memory Retrieval

High-Fidelity Novel View Synthesis via Splatting-Guided Diffusion

Noise May Contain Transferable Knowledge: Understanding Semi-supervised Heterogeneous Domain Adaptation from an Empirical Perspective