Merge pull request #178 from SrWYG/dev0.5.0

[bugfix] IMDB/SST/YELP dataloader char_level无空格 [add]IMDB/SST use spacy tokenizer

reproduction/text_classification/data/IMDBLoader.py

@@ -8,6 +8,7 @@ from fastNLP import Vocabulary
 from fastNLP import Const
 # from reproduction.utils import check_dataloader_paths
 from functools import partial
+from reproduction.utils import check_dataloader_paths, get_tokenizer
 
 
 class IMDBLoader(DataSetLoader):
@@ -22,6 +23,7 @@ class IMDBLoader(DataSetLoader):
 
     def __init__(self):
         super(IMDBLoader, self).__init__()
+        self.tokenizer = get_tokenizer()
 
     def _load(self, path):
         dataset = DataSet()
@@ -32,7 +34,7 @@ class IMDBLoader(DataSetLoader):
                     continue
                 parts = line.split('\t')
                 target = parts[0]
-                words = parts[1].lower().split()
+                words = self.tokenizer(parts[1].lower())
                 dataset.append(Instance(words=words, target=target))
 
         if len(dataset)==0:
@@ -52,13 +54,15 @@ class IMDBLoader(DataSetLoader):
         for name, path in paths.items():
             dataset = self.load(path)
             datasets[name] = dataset
-            
+
         def wordtochar(words):
             chars = []
             for word in words:
                 word = word.lower()
                 for char in word:
                     chars.append(char)
+                chars.append('')
+            chars.pop()
             return chars
 
         if char_level_op:

reproduction/text_classification/data/sstLoader.py

@@ -7,6 +7,7 @@ from fastNLP import Instance
 from fastNLP.io.embed_loader import EmbeddingOption, EmbedLoader
 import csv
 from typing import Union, Dict
+from reproduction.utils import check_dataloader_paths, get_tokenizer
 
 class SSTLoader(DataSetLoader):
     URL = 'https://nlp.stanford.edu/sentiment/trainDevTestTrees_PTB.zip'
@@ -104,6 +105,7 @@ class sst2Loader(DataSetLoader):
     '''
     def __init__(self):
         super(sst2Loader, self).__init__()
+        self.tokenizer = get_tokenizer()
 
     def _load(self, path: str) -> DataSet:
         ds = DataSet()
@@ -114,7 +116,7 @@ class sst2Loader(DataSetLoader):
             if idx<=skip_row:
                 continue
             target = row[1]
-            words = row[0].split()
+            words=self.tokenizer(row[0])
             ds.append(Instance(words=words,target=target))
             all_count+=1
         print("all count:", all_count)
@@ -135,11 +137,13 @@ class sst2Loader(DataSetLoader):
             datasets[name] = dataset
 
         def wordtochar(words):
-            chars=[]
+            chars = []
             for word in words:
-                word=word.lower()
+                word = word.lower()
                 for char in word:
                     chars.append(char)
+                chars.append('')
+            chars.pop()
             return chars
 
         input_name, target_name = 'words', 'target'

reproduction/text_classification/data/yelpLoader.py

@@ -141,14 +141,14 @@ class yelpLoader(DataSetLoader):
         _train_ds = [info.datasets[name]
                      for name in train_ds] if train_ds else info.datasets.values()
 
-
         def wordtochar(words):
-
-            chars=[]
+            chars = []
             for word in words:
-                word=word.lower()
+                word = word.lower()
                 for char in word:
                     chars.append(char)
+                chars.append('')
+            chars.pop()
             return chars
 
         input_name, target_name = 'words', 'target'