A major bug fix for location sensitive attention.

layers/attention.py

@@ -69,10 +69,18 @@ class LocationSensitiveAttention(nn.Module):
 
 
 class AttentionRNN(nn.Module):
-    def __init__(self, out_dim, annot_dim, memory_dim):
+    def __init__(self, out_dim, annot_dim, memory_dim, align_model):
         super(AttentionRNN, self).__init__()
         self.rnn_cell = nn.GRUCell(out_dim + memory_dim, out_dim)
+        # pick bahdanau or location sensitive attention
+        if align_model == 'b':
+            self.alignment_model = BahdanauAttention(annot_dim, out_dim, out_dim)
+        if align_model == 'ls':
             self.alignment_model = LocationSensitiveAttention(annot_dim, out_dim, out_dim)
+        else:
+            raise RuntimeError(" Wrong alignment model name: {}. Use\
+                'b' (Bahdanau) or 'ls' (Location Sensitive).".format(align_model))
+            
 
     def forward(self, memory, context, rnn_state, annotations,
                 attention_vec, mask=None, annotations_lengths=None):
@@ -88,6 +96,9 @@ class AttentionRNN(nn.Module):
         # Alignment
         # (batch, max_time)
         # e_{ij} = a(s_{i-1}, h_j)
+        if attnetion_vec is None:
+            alignment = self.alignment_model(annotations, rnn_output)
+        else:
             alignment = self.alignment_model(annotations, rnn_output, attention_vec)
 
         # TODO: needs recheck.

layers/tacotron.py

@@ -212,12 +212,10 @@ class Decoder(nn.Module):
         eps (float): threshold for detecting the end of a sentence.
     """
 
-    def __init__(self, in_features, memory_dim, r, eps=0, mode='train'):
+    def __init__(self, in_features, memory_dim, r):
         super(Decoder, self).__init__()
-        self.mode = mode
         self.max_decoder_steps = 200
         self.memory_dim = memory_dim
-        self.eps = eps
         self.r = r
         # memory -> |Prenet| -> processed_memory
         self.prenet = Prenet(memory_dim * r, out_features=[256, 128])
@@ -268,8 +266,8 @@ class Decoder(nn.Module):
             for _ in range(len(self.decoder_rnns))]
         current_context_vec = inputs.data.new(B, 256).zero_()
         stopnet_rnn_hidden = inputs.data.new(B, self.r * self.memory_dim).zero_()
-        attention_vec = inputs.data.new(B, T).zero_()
+        attention = inputs.data.new(B, T).zero_()
-        attention_vec_cum = inputs.data.new(B, T).zero_()
+        attention_cum = inputs.data.new(B, T).zero_()
         # Time first (T_decoder, B, memory_dim)
         if memory is not None:
             memory = memory.transpose(0, 1)
@@ -287,12 +285,12 @@ class Decoder(nn.Module):
             # Prenet
             processed_memory = self.prenet(memory_input)
             # Attention RNN
-            attention_vec_cat = torch.cat((attention_vec.unsqueeze(1),
+            attention_cat = torch.cat((attention.unsqueeze(1),
-                                               attention_vec_cum.unsqueeze(1) / (t + 1)),
+                                               attention_cum.unsqueeze(1) / (t + 1)),
                                            dim=1)
             attention_rnn_hidden, current_context_vec, attention = self.attention_rnn(
-                processed_memory, current_context_vec, attention_rnn_hidden, inputs, attention_vec_cat)
+                processed_memory, current_context_vec, attention_rnn_hidden, inputs, attention_cat)
-            attention_vec_cum += attention_vec
+            attention_cum += attention
             # Concat RNN output and attention context vector
             decoder_input = self.project_to_decoder_in(
                 torch.cat((attention_rnn_hidden, current_context_vec), -1))
@@ -333,6 +331,7 @@ class Decoder(nn.Module):
 class StopNet(nn.Module):
     
     def __init__(self, r, memory_dim):
+        """Predicts the stop token to stop the decoder at testing time"""
         super(StopNet, self).__init__()
         self.rnn = nn.GRUCell(memory_dim * r, memory_dim * r)
         self.relu = nn.ReLU()