torchcell.models.hetero_cell_bipartite_dango_gi.create_conv_layer

create_conv_layer(encoder_type: str, in_channels: int, out_channels: int, config: dict[str, Any], edge_dim: int | None = None, dropout: float = 0.1) → Module

Create appropriate conv layer based on encoder type.

Parameters:

• encoder_type – Type of encoder - “gatv2” or “gin”

• in_channels – Input channel dimension

• out_channels – Output channel dimension

• config – Configuration dict for the encoder

• edge_dim – Edge feature dimension (for GATv2)

• dropout – Dropout rate for GIN MLP

Returns:

Conv layer (GATv2Conv or GINConv)