UNIST-Eunchan
commited on
Commit
•
56b6b47
1
Parent(s):
ac0ecce
Update app.py
Browse files
app.py
CHANGED
@@ -36,6 +36,10 @@ def infer(input_ids, max_length, temperature, top_k, top_p):
|
|
36 |
def chunking(book_text):
|
37 |
sentences = sent_tokenize(book_text)
|
38 |
segments = []
|
|
|
|
|
|
|
|
|
39 |
#sentences, token_lens
|
40 |
current_segment = ""
|
41 |
total_token_lens = 0
|
|
|
36 |
def chunking(book_text):
|
37 |
sentences = sent_tokenize(book_text)
|
38 |
segments = []
|
39 |
+
token_lens = []
|
40 |
+
|
41 |
+
for sent_i_th in sentences:
|
42 |
+
token_lens.append(len(tokenizer.tokenize(sent_i_th)))
|
43 |
#sentences, token_lens
|
44 |
current_segment = ""
|
45 |
total_token_lens = 0
|