Flow Model
A class of generative models that model the evolution of probability densities through continuous-time ordinary differential equations, learning a velocity field from a source distribution to a target distribution.
Timeline (last 90 days)
The scalability of flow models to long contexts and complex reasoning tasks is a direction to explore in the future.
Diffusion models and flow models generate samples through iterative steps, and each step requires a full forward pass through a large network, a process that dominates the generation cost.
Diffusion models, flow models and SB models are the three mainstream frameworks of current generative models.
All Facts (3)
Diffusion models and flow models generate samples through iterative steps, and each step requires a full forward pass through a large network, a process that dominates the generation cost.
50%UnverifiedDiffusion models, flow models and SB models are the three mainstream frameworks of current generative models.
50%UnverifiedThe scalability of flow models to long contexts and complex reasoning tasks is a direction to explore in the future.
50%