NoiseSchedule class

Constructors

NoiseSchedule.linear({int numTimesteps = 1000, double betaStart = 1e-4, double betaEnd = 0.02})
Ho-et-al linear beta schedule.
factory

Properties

alphaBars → Float32List
final
alphas → Float32List
final
betas → Float32List
final
hashCode → int
The hash code for this object.
no setterinherited
numTimesteps → int
final
posteriorVariance → Float32List
final
runtimeType → Type
A representation of the runtime type of the object.
no setterinherited
sqrtAlphaBars → Float32List
final
sqrtOneMinusAlphaBars → Float32List
final
sqrtRecipAlphas → Float32List
final

Methods

forwardDiffuse(Tensor x0, int t, {int? seed, Tensor? eps}) → ({Tensor eps, Tensor xT})
Closed-form forward diffusion: x_t = √α̅_t · x_0 + √(1 - α̅_t) · ε with ε ~ N(0, I). Returns the noised sample and the noise it used (so training code can regress ε̂ → ε). Pass eps to make the noise deterministic; otherwise it is drawn using seed (defaults to the system RNG).
noSuchMethod(Invocation invocation) → dynamic
Invoked when a nonexistent method or property is accessed.
inherited
reverseStep(Tensor xt, int t, Tensor epsHat, {int? seed}) → Tensor
One Langevin step of the reverse chain given a predicted noise epsHat (same shape as xt).
toString() → String
A string representation of this object.
inherited

Operators

operator ==(Object other) → bool
The equality operator.
inherited