Example: Excitatory-Inhibitory RNN
import nn4n
from nn4n.model import CTRNN
rnn = CTRNN()
rnn.print_layers()
This will create a Vanilla RNN with default parameters. The printed details of the network will be:
Linear Layer:
| input_dim: 1
| output_dim: 100
| weight_learnable: True
| weight_min: -0.9696310758590698
| weight_max: 0.9939578771591187
| bias_learnable: False
| bias_min: 0.0
| bias_max: 0.0
| sparsity: 1.0
Recurrence:
| init_hidden_min: 0.0
| init_hidden_max: 0.0
| preact_noise: 0
| postact_noise: 0
| activation: relu
| alpha: 0.1
| init_state: zero
| init_state_learnable: False
Hidden Layer:
| input_dim: 100
| output_dim: 100
| weight_learnable: True
| weight_min: -0.09999535232782364
| weight_max: 0.09997699409723282
| bias_learnable: False
| bias_min: 0.0
| bias_max: 0.0
| sparsity: 1
Linear Layer:
| input_dim: 100
| output_dim: 1
| weight_learnable: True
| weight_min: -0.09923813492059708
| weight_max: 0.09776326268911362
| bias_learnable: False
| bias_min: 0.0
| bias_max: 0.0
| sparsity: 1.0