AI研究論文デイリー

Continuous Speech Synthesis using per-token Latent Diffusion

Hybrid Preferences: Learning to Route Instances for Human vs. AI Feedback

Reflection-Bench: probing AI intelligence with reflection

Leveraging Skills from Unlabeled Prior Data for Efficient Online Exploration

Are LLMs Better than Reported? Detecting Label Errors and Mitigating Their Effect on Model Performance

Infinity-MM: Scaling Multimodal Performance with Large-Scale and High-Quality Instruction Data

Counting Ability of Large Language Models and Impact of Tokenization

FasterCache: Training-Free Video Diffusion Model Acceleration with High Quality

Analysing the Residual Stream of Language Models Under Knowledge Conflicts

Dynamic 3D Gaussian Tracking for Graph-Based Neural Dynamics Modeling

MMAU: A Massive Multi-Task Audio Understanding and Reasoning Benchmark

ROCKET-1: Master Open-World Interaction with Visual-Temporal Context Prompting

Fictitious Synthetic Data Can Improve LLM Factuality via Prerequisite Learning

Teach Multimodal LLMs to Comprehend Electrocardiographic Images

Mapping the Media Landscape: Predicting Factual Reporting and Political Bias Through Web Interactions

Read-ME: Refactorizing LLMs as Router-Decoupled Mixture of Experts with System Co-Design