jaxsnn.discrete.modules.linear

Classes

DenseData	alias of jax.Array
Parameter	alias of jax.Array
Projection(generator, int], …)	A projection module for the discrete paradigm.

Functions

jaxsnn.discrete.modules.linear.Linear(mean: float = 0.5, std: float = 2.0, pre_weights: Optional[jax.Array, None] = None) → jaxsnn.discrete.types.Projection

Creates a linear projection layer

Either:

• initialize weights from a Gaussian (mean, std), or

• provide a concrete weight array.

Parameters

• mean – Mean value for weight initialization.

• std – Standard deviation for weight initialization.

• pre_weights – Optional weight array. If provided, mean and std are ignored.

Returns

A Projection object containing the layer definition.

jaxsnn.discrete.modules.linear.linear(inputs: Dict[str, jax.Array], state: None, weight: jax.Array) → Tuple[None, jax.Array]