controllability
The degree to which an AI system can be directed or guided by users to produce desired outcomes or behaviors in its outputs.
- A Controllable Examination for Long-Context Language Models
- CCS: Controllable and Constrained Sampling with Diffusion Models via Initial Noise Perturbation
- DISCO: DISCrete nOise for Conditional Control in Text-to-Image Diffusion Models
- Frame In-N-Out: Unbounded Controllable Image-to-Video Generation
- Image as a World: Generating Interactive World from Single Image via Panoramic Video Generation
- LLM Unlearning via Neural Activation Redirection
- LaViDa: A Large Diffusion Model for Vision-Language Understanding
- Localizing Knowledge in Diffusion Transformers
- Measuring and Guiding Monosemanticity
- Remasking Discrete Diffusion Models with Inference-Time Scaling
- Reward-Instruct: A Reward-Centric Approach to Fast Photo-Realistic Image Generation
- Treasure Hunt: Real-time Targeting of the Long Tail using Training-Time Markers