state of the art
The highest level of development achieved in a particular field of AI, benchmarked against competitors or previous methods.
- ALMGuard: Safety Shortcuts and Where to Find Them as Guardrails for Audio–Language Models
- ATLAS: Autoformalizing Theorems through Lifting, Augmentation, and Synthesis of Data
- Accelerated Sampling from Masked Diffusion Models via Entropy Bounded Unmasking
- Breaking the Batch Barrier (B3) of Contrastive Learning via Smart Batch Mining
- Efficient $k$-Sparse Band–Limited Interpolation with Improved Approximation Ratio
- Instance-Level Composed Image Retrieval
- Learning Skill-Attributes for Transferable Assessment in Video
- NerfBaselines: Consistent and Reproducible Evaluation of Novel View Synthesis Methods
- Optimal Online Change Detection via Random Fourier Features
- Part-Aware Bottom-Up Group Reasoning for Fine-Grained Social Interaction Detection
- Position: Bridge the Gaps between Machine Unlearning and AI Regulation
- Rethinking Residual Distribution in Locate-then-Edit Model Editing
- Tiled Flash Linear Attention: More Efficient Linear RNN and xLSTM Kernels
- VIPAMIN: Visual Prompt Initialization via Embedding Selection and Subspace Expansion
- Whose Instructions Count? Resolving Preference Bias in Instruction Fine-Tuning