# CPU recurrent regression network for the Python API examples.
LSTM:
  activation: tanh
  activity_regularization: l2
  activityregularizer_value: 0.0
  bidirectional: false
  kernel_regularization: l2
  kernelregularizer_value: 0.0
  lstm_bn: false
  nlayers: 1
  type: GRU
  units: 32
dense:
  activation: [relu]
  dense_bn: [false]
  dropout: [0.0]
  kernel_regularizer: [l2]
  regularization: [false]
  regularizer_value: [0.0]
  neurons: [32]
  nlayers: 1
embedding:
  length: 16
learning:
  batch_size: 64
  epochs: 150
  learningrate: 0.003
  optimizer: adamw
  weight_decay: 0.01
  device: cpu
  seed: 42
  deterministic: true
output:
  rp-activation: linear
  rp-column: rp
  rp-dimension: 1
  rp-loss: mse
  rp-metrics: mse
  rp-weight: 1
siamese:
  use: false
  merge_type: add
  single_predictions: false
callbacks:
  check_point: false
  log_csv: false
  early_stopping: true
  early_stopping_patience: 20
  tensor_board: false
  progressbar: true
  reduce_lr: true
  reduce_lr_factor: 0.5
  reduce_lr_patience: 8
predictions:
  continues: [rp]
  fractions: []
encoder:
  route: recurrent
tracking:
  enabled: false