update

train_eval.py

@@ -8,6 +8,7 @@ import time
 from utils import get_time_dif
 
 
+# 权重初始化，默认xavier
 def init_network(model, method='xavier', exclude='embedding', seed=123):
     for name, w in model.named_parameters():
         if exclude not in name:
@@ -20,17 +21,16 @@ def init_network(model, method='xavier', exclude='embedding', seed=123):
                     nn.init.normal_(w)
             elif 'bias' in name:
                 nn.init.constant_(w, 0)
-            else: 
+            else:
                 pass
 
 
 def train(config, model, train_iter, dev_iter, test_iter):
     start_time = time.time()
     model.train()
-    # criterion = nn.CrossEntropyLoss()
     optimizer = torch.optim.Adam(model.parameters(), lr=config.learning_rate)
 
-    # 学习率指数衰减，每次epoch：学习率 ×= gamma 
+    # 学习率指数衰减，每次epoch：学习率 = gamma * 学习率
     # scheduler = torch.optim.lr_scheduler.ExponentialLR(optimizer, gamma=0.9)
     total_batch = 0  # 记录进行到多少batch
     dev_best_loss = float('inf')
@@ -43,10 +43,7 @@ def train(config, model, train_iter, dev_iter, test_iter):
         for i, (trains, labels) in enumerate(train_iter):
             outputs = model(trains)
             model.zero_grad()
-            # print(outputs.size())
-            # print(labels.size())
             loss = F.cross_entropy(outputs, labels)
-            # loss = criterion(outputs, labels)
             loss.backward()
             optimizer.step()
             if total_batch % 100 == 0: 

utils_fasttext.py

@@ -133,7 +133,6 @@ class DatasetIterater(object):
             return self.n_batches + 1
         else:
             return self.n_batches
-        
 
 
 def build_iterator(dataset, config):
@@ -146,3 +145,23 @@ def get_time_dif(start_time):
     end_time = time.time()
     time_dif = end_time - start_time
     return timedelta(seconds=int(round(time_dif)))
+
+if __name__ == "__main__":
+    '''提取预训练词向量'''
+    vocab_dir = "./THUCNews/data/vocab.pkl"
+    pretrain_dir = "./THUCNews/data/sgns.sogou.char"
+    emb_dim = 300
+    filename_trimmed_dir = "./THUCNews/data/vocab.embedding.sougou"
+    word_to_id = pkl.load(open(vocab_dir, 'rb'))
+    embeddings = np.random.rand(len(word_to_id), emb_dim)
+    f = open(pretrain_dir, "r", encoding='UTF-8')
+    for i, line in enumerate(f.readlines()):
+        # if i == 0:  # 若第一行是标题，则跳过
+        #     continue
+        lin = line.strip().split(" ")
+        if lin[0] in word_to_id:
+            idx = word_to_id[lin[0]]
+            emb = [float(x) for x in lin[1:301]]
+            embeddings[idx] = np.asarray(emb, dtype='float32')
+    f.close()
+    np.savez_compressed(filename_trimmed_dir, embeddings=embeddings)