AI研究论文每日精选

DeepSeek-Coder-V2: Breaking the Barrier of Closed-Source Models in Code Intelligence

Depth Anywhere: Enhancing 360 Monocular Depth Estimation via Perspective Distillation and Unlabeled Data Augmentation

Bootstrapping Language Models with DPO Implicit Rewards

TroL: Traversal of Layers for Large Language and Vision Models

ChatGLM: A Family of Large Language Models from GLM-130B to GLM-4 All Tools

VoCo-LLaMA: Towards Vision Compression with Large Language Models

AgileCoder: Dynamic Collaborative Agents for Software Development based on Agile Methodology

From RAGs to rich parameters: Probing how language models utilize external knowledge over parametric information for factual queries

Learn Beyond The Answer: Training Language Models with Reflection for Mathematical Reasoning

SafeInfer: Context Adaptive Decoding Time Safety Alignment for Large Language Models

RepLiQA: A Question-Answering Dataset for Benchmarking LLMs on Unseen Reference Content

Safety Arithmetic: A Framework for Test-time Safety Alignment of Language Models by Steering Parameters and Activations

Tokenization Falling Short: The Curse of Tokenization

Benchmarking Multi-Image Understanding in Vision and Language Models: Perception, Knowledge, Reasoning, and Multi-Hop Reasoning

OlympicArena: Benchmarking Multi-discipline Cognitive Reasoning for Superintelligent AI

HumanSplat: Generalizable Single-Image Human Gaussian Splatting with Structure Priors

Large Scale Transfer Learning for Tabular Data via Language Modeling

Estimating Knowledge in Large Language Models Without Generating a Single Token

Language Models are Surprisingly Fragile to Drug Names in Biomedical Benchmarks

Not All Prompts Are Made Equal: Prompt-based Pruning of Text-to-Image Diffusion Models

Mixture of Scales: Memory-Efficient Token-Adaptive Binarization for Large Language Models

BPO: Supercharging Online Preference Learning by Adhering to the Proximity of Behavior LLM

From Crowdsourced Data to High-Quality Benchmarks: Arena-Hard and BenchBuilder Pipeline

VIA: A Spatiotemporal Video Adaptation Framework for Global and Local Video Editing

Hierarchical Prompting Taxonomy: A Universal Evaluation Framework for Large Language Models

Adversarial Attacks on Multimodal Agents

Immiscible Diffusion: Accelerating Diffusion Training with Noise Assignment

JEN-1 DreamStyler: Customized Musical Concept Learning via Pivotal Parameters Tuning

Super(ficial)-alignment: Strong Models May Deceive Weak Models in Weak-to-Strong Generalization

Mixture-of-Subspaces in Low-Rank Adaptation