Merge pull request #84 from neulab/decoder-dim

setting decoder RNN input dim to size of trg embeddings, not of encoder output

xnmt/decoder.py

@@ -31,7 +31,7 @@ def rnn_from_spec(spec, num_layers, input_dim, hidden_dim, model, residual_to_ou
 class MlpSoftmaxDecoder(Decoder):
   # TODO: This should probably take a softmax object, which can be normal or class-factored, etc.
   # For now the default behavior is hard coded.
-  def __init__(self, layers, input_dim, lstm_dim, mlp_hidden_dim, vocab_size, model,
+  def __init__(self, layers, input_dim, lstm_dim, mlp_hidden_dim, vocab_size, model, trg_embed_dim,
                fwd_lstm=None):
     self.input_dim = input_dim
     if fwd_lstm == None:
@@ -40,7 +40,7 @@ def __init__(self, layers, input_dim, lstm_dim, mlp_hidden_dim, vocab_size, mode
       self.fwd_lstm = fwd_lstm
     self.mlp = MLP(input_dim + lstm_dim, mlp_hidden_dim, vocab_size, model)
     self.state = None
-    self.serialize_params = [layers, input_dim, lstm_dim, mlp_hidden_dim, vocab_size, model]
+    self.serialize_params = [layers, input_dim, lstm_dim, mlp_hidden_dim, vocab_size, model, trg_embed_dim]
 
   def initialize(self):
     self.state = self.fwd_lstm.initial_state()

xnmt/xnmt_train.py

@@ -157,11 +157,11 @@ def create_model(self):
     self.attender = StandardAttender(self.attention_context_dim, self.output_state_dim, self.attender_hidden_dim,
                                      self.model)
 
-    decoder_rnn = Decoder.rnn_from_spec(self.args.decoder_type, self.args.decoder_layers, self.attention_context_dim,
+    decoder_rnn = Decoder.rnn_from_spec(self.args.decoder_type, self.args.decoder_layers, self.output_word_emb_dim,
                                         self.output_state_dim, self.model, self.args.residual_to_output)
     self.decoder = MlpSoftmaxDecoder(self.args.decoder_layers, self.attention_context_dim, self.output_state_dim,
                                      self.output_mlp_hidden_dim, len(self.output_reader.vocab),
-                                     self.model, decoder_rnn)
+                                     self.model, self.output_word_emb_dim, decoder_rnn)
 
     self.translator = DefaultTranslator(self.input_embedder, self.encoder, self.attender, self.output_embedder, self.decoder)
     self.model_params = ModelParams(self.encoder, self.attender, self.decoder, self.input_reader.vocab.i2w,