data diversity
Data diversity refers to the variety and richness of data samples across different dimensions, such as classes, modalities, or sources. In AI, ensuring data diversity is crucial for developing robust models that can perform well across a range of real-world scenarios.
- Beyond Token Probes: Hallucination Detection via Activation Tensors with ACT-ViT
- CCS: Controllable and Constrained Sampling with Diffusion Models via Initial Noise Perturbation
- CoIDO: Efficient Data Selection for Visual Instruction Tuning via Coupled Importance-Diversity Optimization
- DexFlyWheel: A Scalable and Self-improving Data Generation Framework for Dexterous Manipulation
- Diversity as a Reward: Fine-Tuning LLMs on a Mixture of Domain-Undetermined Data
- Learning from Reward-Free Offline Data: A Case for Planning with Latent Dynamics Models
- PANGEA: Projection-Based Augmentation with Non-Relevant General Data for Enhanced Domain Adaptation in LLMs
- Prismatic Synthesis: Gradient-based Data Diversification Boosts Generalization in LLM Reasoning
- RigAnyFace: Scaling Neural Facial Mesh Auto-Rigging with Unlabeled Data
- STAR: A Benchmark for Astronomical Star Fields Super-Resolution
- TreeSynth: Synthesizing Diverse Data from Scratch via Tree-Guided Subspace Partitioning