lib sync

d2l-ai · Oct 23, 2020 · e90d41b · e90d41b
1 parent 54f2725
commit e90d41b
Show file tree

Hide file tree

Showing 3 changed files with 8 additions and 8 deletions.
diff --git a/d2l/mxnet.py b/d2l/mxnet.py
@@ -840,9 +840,9 @@ def load_data_nmt(batch_size, num_steps, num_examples=600):
     """Return the iterator and the vocabularies of the translation dataset."""
     text = preprocess_nmt(read_data_nmt())
     source, target = tokenize_nmt(text, num_examples)
-    src_vocab = d2l.Vocab(source, min_freq=2, 
+    src_vocab = d2l.Vocab(source, min_freq=2,
                           reserved_tokens=['<pad>', '<bos>', '<eos>'])
-    tgt_vocab = d2l.Vocab(target, min_freq=2, 
+    tgt_vocab = d2l.Vocab(target, min_freq=2,
                           reserved_tokens=['<pad>', '<bos>', '<eos>'])
     src_array, src_valid_len = build_array_nmt(source, src_vocab, num_steps)
     tgt_array, tgt_valid_len = build_array_nmt(target, tgt_vocab, num_steps)
@@ -976,7 +976,7 @@ def predict_s2s_ch9(model, src_sentence, src_vocab, tgt_vocab, num_steps,
         # of the decoder at the next time step
         dec_X = Y.argmax(axis=2)
         pred = dec_X.squeeze(axis=0).astype('int32').item()
-        # Once the end-of-sequence token is predicted, the generation of 
+        # Once the end-of-sequence token is predicted, the generation of
         # the output sequence is complete
         if pred == tgt_vocab['<eos>']:
             break

diff --git a/d2l/tensorflow.py b/d2l/tensorflow.py
@@ -843,9 +843,9 @@ def load_data_nmt(batch_size, num_steps, num_examples=600):
     """Return the iterator and the vocabularies of the translation dataset."""
     text = preprocess_nmt(read_data_nmt())
     source, target = tokenize_nmt(text, num_examples)
-    src_vocab = d2l.Vocab(source, min_freq=2, 
+    src_vocab = d2l.Vocab(source, min_freq=2,
                           reserved_tokens=['<pad>', '<bos>', '<eos>'])
-    tgt_vocab = d2l.Vocab(target, min_freq=2, 
+    tgt_vocab = d2l.Vocab(target, min_freq=2,
                           reserved_tokens=['<pad>', '<bos>', '<eos>'])
     src_array, src_valid_len = build_array_nmt(source, src_vocab, num_steps)
     tgt_array, tgt_valid_len = build_array_nmt(target, tgt_vocab, num_steps)

diff --git a/d2l/torch.py b/d2l/torch.py
@@ -903,9 +903,9 @@ def load_data_nmt(batch_size, num_steps, num_examples=600):
     """Return the iterator and the vocabularies of the translation dataset."""
     text = preprocess_nmt(read_data_nmt())
     source, target = tokenize_nmt(text, num_examples)
-    src_vocab = d2l.Vocab(source, min_freq=2, 
+    src_vocab = d2l.Vocab(source, min_freq=2,
                           reserved_tokens=['<pad>', '<bos>', '<eos>'])
-    tgt_vocab = d2l.Vocab(target, min_freq=2, 
+    tgt_vocab = d2l.Vocab(target, min_freq=2,
                           reserved_tokens=['<pad>', '<bos>', '<eos>'])
     src_array, src_valid_len = build_array_nmt(source, src_vocab, num_steps)
     tgt_array, tgt_valid_len = build_array_nmt(target, tgt_vocab, num_steps)
@@ -1062,7 +1062,7 @@ def predict_s2s_ch9(model, src_sentence, src_vocab, tgt_vocab, num_steps,
         # of the decoder at the next time step
         dec_X = Y.argmax(dim=2)
         pred = dec_X.squeeze(dim=0).type(torch.int32).item()
-        # Once the end-of-sequence token is predicted, the generation of 
+        # Once the end-of-sequence token is predicted, the generation of
         # the output sequence is complete
         if pred == tgt_vocab['<eos>']:
             break