Flow Mapping
A distillation form of continuous flow models that learns a deterministic transformation from noise to data, enabling single-step or few-step generation and substantially reducing inference latency.
Core Facts
Timeline (last 90 days)
Lafa's flow model can be distilled into flow maps, enabling single-step or few-step generation and thereby reducing inference latency.
Flow matching models generate images by learning a continuous probability flow (ODE trajectory) from the noise distribution to the data distribution, and are generally more stable to train and require fewer inference steps.
All Facts (2)
Flow matching models generate images by learning a continuous probability flow (ODE trajectory) from the noise distribution to the data distribution, and are generally more stable to train and require fewer inference steps.
75%UnverifiedLafa's flow model can be distilled into flow maps, enabling single-step or few-step generation and thereby reducing inference latency.
50%