add padding idx for emb and delete useless code in reader

dygraph/seq2seq/args.py

@@ -56,7 +56,7 @@ def parse_args():
         "--padding_idx",
         type=int,
         default=0,
-        help="padding_idx of embedding")
+        help="padding index of embedding")
 
     parser.add_argument(
         "--num_layers",

dygraph/seq2seq/attention_model.py

@@ -15,7 +15,7 @@
 import numpy as np
 import paddle
 from paddle.nn import Layer, Linear, Dropout, Embedding, LayerList, RNN, LSTM, LSTMCell, RNNCellBase
-from paddle.nn.initializer import Uniform
+import paddle.nn.initializer as I
 import paddle.nn.functional as F
 SEED = 102
 paddle.framework.manual_seed(SEED)
@@ -84,9 +84,10 @@ class Encoder(Layer):
         self.embedder = Embedding(
             vocab_size,
             hidden_size,
+            padding_idx=padding_idx,
             weight_attr=paddle.ParamAttr(
                 name='source_embedding',
-                initializer=Uniform(
+                initializer=I.Uniform(
                     low=-init_scale, high=init_scale)))
         self.lstm = LSTM(
             input_size=hidden_size,
@@ -111,13 +112,13 @@ class AttentionLayer(Layer):
         self.input_proj = Linear(
             hidden_size,
             hidden_size,
-            weight_attr=paddle.ParamAttr(initializer=Uniform(
+            weight_attr=paddle.ParamAttr(initializer=I.Uniform(
                 low=-init_scale, high=init_scale)),
             bias_attr=bias)
         self.output_proj = Linear(
             hidden_size + hidden_size,
             hidden_size,
-            weight_attr=paddle.ParamAttr(initializer=Uniform(
+            weight_attr=paddle.ParamAttr(initializer=I.Uniform(
                 low=-init_scale, high=init_scale)),
             bias_attr=bias)
 
@@ -202,9 +203,10 @@ class Decoder(Layer):
         self.embedder = Embedding(
             vocab_size,
             hidden_size,
+            padding_idx=padding_idx,
             weight_attr=paddle.ParamAttr(
                 name='target_embedding',
-                initializer=Uniform(
+                initializer=I.Uniform(
                     low=-init_scale, high=init_scale)))
         self.dropout = dropout
         self.lstm_attention = RNN(DecoderCell(hidden_size, hidden_size,
@@ -214,7 +216,7 @@ class Decoder(Layer):
         self.fc = Linear(
             hidden_size,
             vocab_size,
-            weight_attr=paddle.ParamAttr(initializer=Uniform(
+            weight_attr=paddle.ParamAttr(initializer=I.Uniform(
                 low=-init_scale, high=init_scale)),
             bias_attr=False)
 

dygraph/seq2seq/data

+/models/dygraph/seq2seq_attn/data
\ No newline at end of file

dygraph/seq2seq/reader.py

@@ -177,17 +177,6 @@ class IWSLTDataset(Dataset):
     def __len__(self):
         return self.num_samples
 
-    def get_max_seq_len(self):
-        src_max_seq_len = 0
-        trg_max_seq_len = 0
-        for data in self.src_data:
-            src_max_seq_len = max(src_max_seq_len, len(data))
-        for data in self.tar_data:
-            trg_max_seq_len = max(trg_max_seq_len, len(data))
-        src_max_seq_len = min(src_max_seq_len, self.max_seq_len)
-        trg_max_seq_len = min(trg_max_seq_len, self.max_seq_len)
-        return src_max_seq_len, trg_max_seq_len
-
 
 class DataCollector():
     def __init__(self):