Articles de Recherche en IA Quotidiens

Breaking the Memory Barrier: Near Infinite Batch Size Scaling for Contrastive Loss

Can Knowledge Editing Really Correct Hallucinations?

LOGO -- Long cOntext aliGnment via efficient preference Optimization

Unleashing Reasoning Capability of LLMs via Scalable Question Synthesis from Scratch

Framer: Interactive Frame Interpolation

Unbounded: A Generative Infinite Game of Character Life Simulation

Steering Knowledge Selection Behaviours in LLMs via SAE-Based Representation Engineering

Distill Visual Chart Reasoning Ability from LLMs to MLLMs

Taipan: Efficient and Expressive State Space Language Models with Selective Attention

Why Does the Effective Context Length of LLMs Fall Short?

SMITE: Segment Me In TimE

MotionCLR: Motion Generation and Training-free Editing via Understanding Attention Mechanisms

Skywork-Reward: Bag of Tricks for Reward Modeling in LLMs

WAFFLE: Multi-Modal Model for Automated Front-End Development

Robust Watermarking Using Generative Priors Against Image Editing: From Benchmarking to Advances

Stable Consistency Tuning: Understanding and Improving Consistency Models

DeCoRe: Decoding by Contrasting Retrieval Heads to Mitigate Hallucinations

CAMEL-Bench: A Comprehensive Arabic LMM Benchmark

CCI3.0-HQ: a large-scale Chinese dataset of high quality designed for pre-training large language models

ADEM-VL: Adaptive and Embedded Fusion for Efficient Vision-Language Tuning

Value Residual Learning For Alleviating Attention Concentration In Transformers

Language Models are Symbolic Learners in Arithmetic

The Nature of Mathematical Modeling and Probabilistic Optimization Engineering in Generative AI

Asynchronous RLHF: Faster and More Efficient Off-Policy RL for Language Models

Should We Really Edit Language Models? On the Evaluation of Edited Language Models

ZIP-FIT: Embedding-Free Data Selection via Compression-Based Alignment

Data Scaling Laws in Imitation Learning for Robotic Manipulation

Pantograph: A Machine-to-Machine Interaction Interface for Advanced Theorem Proving, High Level Reasoning, and Data Extraction in Lean 4

Multi-Draft Speculative Sampling: Canonical Architectures and Theoretical Limits