controllable generation
Controllable generation refers to the ability of generative models to produce outputs that can be influenced by specified parameters or constraints. This allows for targeted creation of content, such as adjusting attributes in images or controlling narrative style in text generation.
- Constrained Discrete Diffusion
- Matryoshka Pilot: Learning to Drive Black-Box LLMs with LLMs
- Noise Consistency Training: A Native Approach for One-step Generator in Learning Additional Controls
- SPMDM: Enhancing Masked Diffusion Models through Simplifing Sampling Path
- TalkCuts: A Large-Scale Dataset for Multi-Shot Human Speech Video Generation
- Temporal In‑Context Fine‑Tuning for Versatile Control of Video Diffusion Models
- URLs Help, Topics Guide: Understanding Metadata Utility in LLM Training