AI研究論文每日精選

Revisit Large-Scale Image-Caption Data in Pre-training Multimodal Foundation Models

SageAttention: Accurate 8-Bit Attention for Plug-and-play Inference Acceleration

Depth Pro: Sharp Monocular Metric Depth in Less Than a Second

Video Instruction Tuning With Synthetic Data

Loong: Generating Minute-level Long Videos with Autoregressive Language Models

Contrastive Localized Language-Image Pre-Training

LLaVA-Critic: Learning to Evaluate Multimodal Models

Large Language Models as Markov Chains

Eliminating Oversaturation and Artifacts of High Guidance Scales in Diffusion Models

VinePPO: Unlocking RL Potential For LLM Reasoning Through Refined Credit Assignment

Distilling an End-to-End Voice Assistant Without Instruction Training Data

Contextual Document Embeddings

CLIP-MoE: Towards Building Mixture of Experts for CLIP with Diversified Multiplet Upcycling

Training Language Models on Synthetic Edit Sequences Improves Code Synthesis

L-CiteEval: Do Long-Context Models Truly Leverage Context for Responding?

Open-RAG: Enhanced Retrieval-Augmented Reasoning with Open-Source Large Language Models

Interpreting and Editing Vision-Language Representations to Mitigate Hallucinations

MedVisionLlama: Leveraging Pre-Trained Large Language Model Layers to Enhance Medical Image Segmentation

Improving Autonomous AI Agents with Reflective Tree Search and Self-Learning

MVGS: Multi-view-regulated Gaussian Splatting for Novel View Synthesis

Vinoground: Scrutinizing LMMs over Dense Temporal Reasoning with Short Videos

Intelligence at the Edge of Chaos

Synthio: Augmenting Small-Scale Audio Classification Datasets with Synthetic Data

Learning the Latent Rules of a Game from Data: A Chess Story

Layer Swapping for Zero-Shot Cross-Lingual Transfer in Large Language Models

Robin3D: Improving 3D Large Language Model via Robust Instruction Tuning

SciPrompt: Knowledge-augmented Prompting for Fine-grained Categorization of Scientific Topics