models

A collection of equivariant neural network architectures.

eMLP(in_rep, out_rep, hidden_units[, ...])	Equivariant MLP composed of eLinear layers.
iMLP(in_rep, out_dim, hidden_units[, ...])	Invariant MLP built from an equivariant backbone and invariant pooling.
MLP(in_dim, out_dim, hidden_units[, ...])	Standard baseline MLP with no symmetry constraints.
eTimeCNNEncoder(in_rep, out_rep, ...[, ...])	Equivariant 1D CNN encoder built from channel-equivariant blocks.
TimeCNNEncoder(in_dim, out_dim, ...[, ...])	1D CNN baseline encoder for inputs of shape (N, in_dim, H).
eCondTransformer(in_rep, cond_rep, out_rep, ...)	Equivariant encoder/decoder Transformer with configurable positional attention.
CondTransformer(in_dim, out_dim, cond_dim, ...)	Encoder/decoder Transformer with configurable positional attention.
GenCondRegressor(in_dim, out_dim, cond_dim)	Generative Conditional Regressor module.
eConditionalUnet1D(in_rep, local_cond_rep[, ...])	Equivariant U-Net for 1D signals with global conditioning and FiLM.
ConditionalUnet1D(input_dim[, ...])	A 1D/Time U-Net for predicting conditional score/velocity fields.