Spaces:
Runtime error
Runtime error
Commit
•
2570705
1
Parent(s):
caff3d5
Upload app.py
Browse files
app.py
ADDED
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from pypinyin import pinyin
|
2 |
+
from transformers import M2M100Tokenizer, M2M100ForConditionalGeneration
|
3 |
+
from LAC import LAC
|
4 |
+
import gradio as gr
|
5 |
+
|
6 |
+
model = M2M100ForConditionalGeneration.from_pretrained("facebook/m2m100_1.2B")
|
7 |
+
tokenizer = M2M100Tokenizer.from_pretrained("facebook/m2m100_1.2B")
|
8 |
+
lac = LAC(mode="seg")
|
9 |
+
|
10 |
+
def make_request(chinese_text):
|
11 |
+
tokenizer.src_lang="zh"
|
12 |
+
encoded_zh = tokenizer(chinese_text, return_tensors="pt")
|
13 |
+
generated_tokens = model.generate(**encoded_zh, forced_bos_token_id=tokenizer.get_lang_id("en"))
|
14 |
+
return tokenizer.batch_decode(generated_tokens, skip_special_tokens=True)
|
15 |
+
|
16 |
+
def generatepinyin(input):
|
17 |
+
pinyin_list = pinyin(input)
|
18 |
+
pinyin_string = ""
|
19 |
+
for piece in pinyin_list:
|
20 |
+
pinyin_string = pinyin_string+" "+piece[0]
|
21 |
+
return pinyin_string
|
22 |
+
|
23 |
+
def generate_response(string_to_translate):
|
24 |
+
response = []
|
25 |
+
response.append([string_to_translate,make_request(string_to_translate),generatepinyin(string_to_translate)])
|
26 |
+
segmented_string_list = lac.run(string_to_translate)
|
27 |
+
for piece in segmented_string_list:
|
28 |
+
response.append([piece,make_request(piece),generatepinyin(piece)])
|
29 |
+
return response
|
30 |
+
|
31 |
+
iface = gr.Interface(
|
32 |
+
fn=generate_response,
|
33 |
+
title="English to Chinese",
|
34 |
+
description="English to Chinese with a SOTA model",
|
35 |
+
inputs=gr.inputs.Textbox(lines=5, placeholder="Enter text in Chinese"),
|
36 |
+
outputs="text")
|
37 |
+
|
38 |
+
iface.launch()
|