调整数据集：韵母对应汉字，政史地 物化生

build_corpus.py

@@ -117,7 +117,7 @@ def align2(sent):
             p = pnyns[i]
             if '\u4e00' <= char <= '\u9fa5':
                 # hanzis.extend([char] + ["_"] * (len(p) - 1))
-                hanzis.extend(["_"] * (len(p) - 1) + [char])
+                hanzis.extend(["_"] * (len(p) - 1) + [char])  #韵母对应汉字
             else:
                 for q in p:
                     hanzis.append(q)
@@ -215,9 +215,9 @@ if __name__ == "__main__":
     # with open("./data/voc/yunmu.txt","r",encoding="utf-8") as f:
     #     yunmus=f.readlines()
     #     yunmus=[a.strip() for a in yunmus]
-    ori_dir="./data/math/train/ori"
-    hanzi_dir="./data/math/train/hanzi"
-    pinyin_dir="./data/math/train/pinyin"
+    ori_dir="./data/zsd/train/ori"
+    hanzi_dir="./data/zsd/train/hanzi"
+    pinyin_dir="./data/zsd/train/pinyin"
     for file in os.listdir(ori_dir):
         build_corpus(os.path.join(ori_dir,file),
                     os.path.join(pinyin_dir,file), os.path.join(hanzi_dir,file))