AI研究論文每日精選

DPO-Shift: Shifting the Distribution of Direct Preference Optimization

Ignore the KL Penalty! Boosting Exploration on Critical Tokens to Enhance RL Fine-Tuning

Towards Trustworthy Retrieval Augmented Generation for Large Language Models: A Survey

WorldGUI: Dynamic Testing for Comprehensive Desktop GUI Automation

TextAtlas5M: A Large-scale Dataset for Dense Text Image Generation

LASP-2: Rethinking Sequence Parallelism for Linear Attention and Its Hybrid

Light-A-Video: Training-free Video Relighting via Progressive Light Fusion

TransMLA: Multi-head Latent Attention Is All You Need

PDE-Controller: LLMs for Autoformalization and Reasoning of PDEs

MetaSC: Test-Time Safety Specification Optimization for Language Models

Fino1: On the Transferability of Reasoning Enhanced LLMs to Finance

SARChat-Bench-2M: A Multi-Task Vision-Language Benchmark for SAR Image Interpretation

LLM Modules: Knowledge Transfer from a Large to a Small Model using Enhanced Cross-Attention

Distillation Scaling Laws

Animate Anyone 2: High-Fidelity Character Image Animation with Environment Affordance

BenchMAX: A Comprehensive Multilingual Evaluation Suite for Large Language Models

Mediator: Memory-efficient LLM Merging with Less Parameter Conflicts and Uncertainty Based Routing

Next Block Prediction: Video Generation via Semi-Autoregressive Modeling

CineMaster: A 3D-Aware and Controllable Framework for Cinematic Text-to-Video Generation

NoLiMa: Long-Context Evaluation Beyond Literal Matching

Homeomorphism Prior for False Positive and Negative Problem in Medical Image Dense Contrastive Representation Learning

LLM Pretraining with Continuous Concepts