pyprobound.layers.conv0d

Nonspecific binding layer.

Members are explicitly re-exported in pyprobound.layers.

Non-specific binding is equivalent to a traditional weight matrix with all its values constrained to be identical to each other, so the only thing that can change the output is the length of the sequence, not its identity. It sets a background enrichment level from which a sequence-specific enrichment is optimized. If it is included in an aggregate, it should be listed first.

Classes

Conv0d(nonspecific, input_shape, ...[, ...])	Non-specific '0D convolution' with length biases.
NonSpecific(alphabet[, ignore_length, name])	Non-specific factor, equivalent to a PSAM of size 1 and equal betas.