semantic alignment
The degree to which the output from generative models aligns with the intended meaning or context of the input. High semantic alignment indicates that the generated content honors the underlying semantics of input data.
- Adaptive Divergence Regularized Policy Optimization for Fine-tuning Generative Models
- Adaptive and Multi-scale Affinity Alignment for Hierarchical Contrastive Learning
- Analogy-based Multi-Turn Jailbreak against Large Language Models
- Asymptotics of SGD in Sequence-Single Index Models and Single-Layer Attention Networks
- Bootstrap Your Uncertainty: Adaptive Robust Classification Driven by Optimal-Transport
- CAMILA: Context-Aware Masking for Image Editing with Language Alignment
- CREA: A Collaborative Multi-Agent Framework for Creative Image Editing and Generation
- Conditional Representation Learning for Customized Tasks
- CovMatch: Cross-Covariance Guided Multimodal Dataset Distillation with Trainable Text Encoder
- DAIL: Beyond Task Ambiguity for Language-Conditioned Reinforcement Learning
- DCI: Dual-Conditional Inversion for Boosting Diffusion-Based Image Editing
- Dual Prototype-Enhanced Contrastive Framework for Class-Imbalanced Graph Domain Adaptation
- EditInfinity: Image Editing with Binary-Quantized Generative Models
- EgoExoBench: A Benchmark for First- and Third-person View Video Understanding in MLLMs
- FACE: A General Framework for Mapping Collaborative Filtering Embeddings into LLM Tokens
- Image as a World: Generating Interactive World from Single Image via Panoramic Video Generation
- LOMIA: Label-Only Membership Inference Attacks against Pre-trained Large Vision-Language Models
- Learning a Cross-Modal Schrödinger Bridge for Visual Domain Generalization
- LuxDiT: Lighting Estimation with Video Diffusion Transformer
- OmniZoom: A Universal Plug-and-Play Paradigm for Cross-Device Smooth Zoom Interpolation
- PC-Net: Weakly Supervised Compositional Moment Retrieval via Proposal-Centric Network
- RespoDiff: Dual-Module Bottleneck Transformation for Responsible & Faithful T2I Generation
- STRIDER: Navigation via Instruction-Aligned Structural Decision Space Optimization
- SemCoT: Accelerating Chain-of-Thought Reasoning through Semantically-Aligned Implicit Tokens
- UniGen: Enhanced Training & Test-Time Strategies for Unified Multimodal Understanding and Generation
- UniHG: A Large-scale Universal Heterogeneous Graph Dataset and Benchmark for Representation Learning and Cross-Domain Transferring