Softmax output and seed

nschaetti · Jan 25, 2019 · c3724df · c3724df
1 parent 3186daf
commit c3724df
Showing 1 changed file with 4 additions and 3 deletions.
diff --git a/echotorch/nn/LiESN.py b/echotorch/nn/LiESN.py
@@ -40,7 +40,7 @@ def __init__(self, input_dim, hidden_dim, output_dim, spectral_radius=0.9,
                  bias_scaling=0, input_scaling=1.0, w=None, w_in=None, w_bias=None, sparsity=None,
                  input_set=[1.0, -1.0], w_sparsity=None, nonlin_func=torch.tanh, learning_algo='inv', ridge_param=0.0,
                  leaky_rate=1.0, train_leaky_rate=False, feedbacks=False, wfdb_sparsity=None,
-                 normalize_feedbacks=False):
+                 normalize_feedbacks=False, softmax_output=False, seed=None):
         """
         Constructor
         :param input_dim:
@@ -67,15 +67,16 @@ def __init__(self, input_dim, hidden_dim, output_dim, spectral_radius=0.9,
                                     w=w, w_in=w_in, w_bias=w_bias, sparsity=sparsity, input_set=input_set,
                                     w_sparsity=w_sparsity, nonlin_func=nonlin_func, learning_algo=learning_algo,
                                     ridge_param=ridge_param, create_cell=False, feedbacks=feedbacks,
-                                    wfdb_sparsity=wfdb_sparsity, normalize_feedbacks=normalize_feedbacks)
+                                    wfdb_sparsity=wfdb_sparsity, normalize_feedbacks=normalize_feedbacks,
+                                    softmax_output=softmax_output, seed=seed)
 
         # Recurrent layer
         self.esn_cell = LiESNCell(leaky_rate, train_leaky_rate, input_dim, hidden_dim, spectral_radius=spectral_radius,
                                   bias_scaling=bias_scaling, input_scaling=input_scaling,
                                   w=w, w_in=w_in, w_bias=w_bias, sparsity=sparsity, input_set=input_set,
                                   w_sparsity=w_sparsity, nonlin_func=nonlin_func, feedbacks=feedbacks,
                                   feedbacks_dim=output_dim, wfdb_sparsity=wfdb_sparsity,
-                                  normalize_feedbacks=normalize_feedbacks)
+                                  normalize_feedbacks=normalize_feedbacks, seed=seed)
     # end __init__
 
     ###############################################