Spaces:
Sleeping
Sleeping
utils with vocab prepration
Browse files
utils.py
ADDED
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
def prepare_vocab(file_path="vocab.txt"):
|
2 |
+
"""Returns Vocab size and decode/encode funcs"""
|
3 |
+
with open(file_path, "r", encoding="utf-8") as f:
|
4 |
+
text = f.read()
|
5 |
+
chars = sorted(list(set(text)))
|
6 |
+
|
7 |
+
vocab_size = len(chars)
|
8 |
+
|
9 |
+
str_to_int = {ch: i for i, ch in enumerate(chars)}
|
10 |
+
int_to_ch = {i: ch for i, ch in enumerate(chars)}
|
11 |
+
|
12 |
+
encode = lambda s: [str_to_int[c] for c in s]
|
13 |
+
decode = lambda t: "".join([int_to_ch[n] for n in t])
|
14 |
+
|
15 |
+
return vocab_size, encode, decode
|