ALD-renoise
A sampling strategy in ALD that estimates the clean sample at each step and then re-injects noise from the current marginal distribution. It is simple to implement but may lose dependencies between states.
Timeline (last 90 days)
The ALD paper studies two sampling processes, ALD-renoise and ALD-posterior, which share the same trained noise predictor and are distinguished only by different scalar inputs.
ALD-renoise adopts a strategy of estimating a clean sample and then re-injecting marginal noise, introducing a random reset between each step that may lose the dependency structure between adjacent states.
All Facts (2)
The ALD paper studies two sampling processes, ALD-renoise and ALD-posterior, which share the same trained noise predictor and are distinguished only by different scalar inputs.
50%UnverifiedALD-renoise adopts a strategy of estimating a clean sample and then re-injecting marginal noise, introducing a random reset between each step that may lose the dependency structure between adjacent states.
50%