fix the problem that network config with the rectifier activation fun…

…ction cannot be saved

cmds/run_Extract_Feats.py

@@ -57,6 +57,7 @@
     numpy_rng = numpy.random.RandomState(89677)
     theano_rng = RandomStreams(numpy_rng.randint(2 ** 30))
     cfg = cPickle.load(open(nnet_cfg,'r'))
+    cfg.init_activation()
     model = None
     if cfg.model_type == 'DNN':
         model = DNN(numpy_rng=numpy_rng, theano_rng = theano_rng, cfg = cfg)

cmds2/run_FeatExt_Kaldi.py

@@ -53,6 +53,7 @@
 
     # load network configuration
     cfg = cPickle.load(open(nnet_cfg,'r'))
+    cfg.init_activation()
 
     # set up the model with model config
     log('> ... setting up the model and loading parameters')

io_func/model_io.py

@@ -82,6 +82,9 @@ def _cfg2file(cfg, filename='cfg.out'):
     cfg.lrate = None
     cfg.train_sets = None; cfg.train_xy = None; cfg.train_x = None; cfg.train_y = None
     cfg.valid_sets = None; cfg.valid_xy = None; cfg.valid_x = None; cfg.valid_y = None
+    cfg.activation = None  # saving the rectifier function causes errors; thus we don't save the activation function
+                           # the activation function is initialized from the activation text ("sigmoid") when the network
+                           # configuration is loaded
     with open(filename, "wb") as output:
         cPickle.dump(cfg, output, cPickle.HIGHEST_PROTOCOL)
 

utils/network_config.py

@@ -97,6 +97,10 @@ def init_data_reading_test(self, data_spec):
         dataset, dataset_args = read_data_args(data_spec)
         self.test_sets, self.test_xy, self.test_x, self.test_y = read_dataset(dataset, dataset_args)
 
+    # initialize the activation function
+    def init_activation(self):
+        self.activation = parse_activation(self.activation_text)
+
     def parse_config_common(self, arguments):
         # parse batch_size, momentum and learning rate 
         if arguments.has_key('batch_size'):

utils/rbm_config.py

@@ -63,7 +63,11 @@ def __init__(self):
     def init_data_reading(self, train_data_spec):
         train_dataset, train_dataset_args = read_data_args(train_data_spec)
         self.train_sets, self.train_xy, self.train_x, self.train_y = read_dataset(train_dataset, train_dataset_args)
-
+
+    # initialize the activation function   
+    def init_activation(self):
+        self.activation = parse_activation(self.activation_text)
+
     # parse the arguments to get the values for various variables 
     def parse_config_common(self, arguments):
         if arguments.has_key('gbrbm_learning_rate'):

utils/sda_config.py

@@ -74,6 +74,10 @@ def init_data_reading(self, train_data_spec):
         train_dataset, train_dataset_args = read_data_args(train_data_spec)
         self.train_sets, self.train_xy, self.train_x, self.train_y = read_dataset(train_dataset, train_dataset_args)
 
+    # initialize the activation function
+    def init_activation(self):
+        self.activation = parse_activation(self.activation_text)
+
     # parse the arguments to get the values for various variables 
     def parse_config_common(self, arguments):
         if arguments.has_key('corruption_level'):