Update by following reviewer's comments.

PaddlePaddle · Sep 12, 2017 · 5510dad · 5510dad
1 parent 69b4bda
commit 5510dad
Show file tree

Hide file tree

Showing 4 changed files with 15 additions and 15 deletions.
diff --git a/mt_with_external_memory/external_memory.py b/mt_with_external_memory/external_memory.py
@@ -56,10 +56,7 @@ def __init__(self,
         self.readonly = readonly
         self.enable_interpolation = enable_interpolation
         self.external_memory = paddle.layer.memory(
-            name=self.name,
-            size=self.mem_slot_size,
-            is_seq=True,
-            boot_layer=boot_layer)
+            name=self.name, size=self.mem_slot_size, boot_layer=boot_layer)
         # prepare a constant (zero) intializer for addressing weights 
         self.zero_addressing_init = paddle.layer.slope_intercept(
             input=paddle.layer.fc(input=boot_layer, size=1),
@@ -114,7 +111,6 @@ def _interpolation(self, head_name, key_vector, addressing_weight):
         last_addressing_weight = paddle.layer.memory(
             name=self.name + "_addressing_weight_" + head_name,
             size=1,
-            is_seq=True,
             boot_layer=self.zero_addressing_init)
         interpolated_weight = paddle.layer.interpolation(
             name=self.name + "_addressing_weight_" + head_name,

diff --git a/mt_with_external_memory/infer.py b/mt_with_external_memory/infer.py
@@ -5,9 +5,10 @@
 import argparse
 import gzip
 import paddle.v2 as paddle
+
 from external_memory import ExternalMemory
-from model import *
-from data_utils import *
+from model import memory_enhanced_seq2seq
+from data_utils import reader_append_wrapper
 
 parser = argparse.ArgumentParser(description=__doc__)
 parser.add_argument(

diff --git a/mt_with_external_memory/model.py b/mt_with_external_memory/model.py
@@ -184,7 +184,7 @@ def recurrent_decoder_step(cur_embedding):
             input=decoder_result, label=target)
         return cost
     else:
-        target_embeddings = paddle.layer.GeneratedInputV2(
+        target_embeddings = paddle.layer.GeneratedInput(
             size=dict_size,
             embedding_name="_decoder_word_embedding",
             embedding_size=word_vec_dim)

diff --git a/mt_with_external_memory/train.py b/mt_with_external_memory/train.py
@@ -6,10 +6,11 @@
 import gzip
 import distutils.util
 import random
+
 import paddle.v2 as paddle
 from external_memory import ExternalMemory
-from model import *
-from data_utils import *
+from model import memory_enhanced_seq2seq
+from data_utils import reader_append_wrapper
 
 parser = argparse.ArgumentParser(description=__doc__)
 parser.add_argument(
@@ -65,6 +66,12 @@ def train():
     """
     For training.
     """
+    # create optimizer
+    optimizer = paddle.optimizer.Adam(
+        learning_rate=5e-5,
+        gradient_clipping_threshold=5,
+        regularization=paddle.optimizer.L2Regularization(rate=8e-4))
+
     # create network config
     source_words = paddle.layer.data(
         name="source_words",
@@ -85,12 +92,8 @@ def train():
         is_generating=False,
         beam_size=None)
 
-    # create parameters and optimizer
+    # create parameters and trainer
     parameters = paddle.parameters.create(cost)
-    optimizer = paddle.optimizer.Adam(
-        learning_rate=5e-5,
-        gradient_clipping_threshold=5,
-        regularization=paddle.optimizer.L2Regularization(rate=8e-4))
     trainer = paddle.trainer.SGD(
         cost=cost, parameters=parameters, update_equation=optimizer)