Spaces:
Sleeping
Sleeping
change token for src lang
Browse files
app.py
CHANGED
@@ -1,18 +1,25 @@
|
|
1 |
import gradio as gr
|
2 |
-
from transformers import
|
3 |
|
4 |
model_name = "dsfsi/nr-en-m2m100-gov"
|
5 |
tokenizer = M2M100Tokenizer.from_pretrained(model_name)
|
|
|
6 |
|
7 |
-
|
8 |
|
9 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
10 |
|
11 |
def translate(inp):
|
12 |
-
# Translate from isiNdebele to English
|
13 |
res = translater_nr_en(inp, max_length=512, early_stopping=True)[0]['translation_text']
|
14 |
return res
|
15 |
|
|
|
16 |
description = """
|
17 |
<p>
|
18 |
<center>
|
@@ -38,3 +45,4 @@ iface = gr.Interface(
|
|
38 |
)
|
39 |
|
40 |
iface.launch(enable_queue=True)
|
|
|
|
1 |
import gradio as gr
|
2 |
+
from transformers import M2M100Tokenizer, M2M100ForConditionalGeneration, pipeline
|
3 |
|
4 |
model_name = "dsfsi/nr-en-m2m100-gov"
|
5 |
tokenizer = M2M100Tokenizer.from_pretrained(model_name)
|
6 |
+
model = M2M100ForConditionalGeneration.from_pretrained(model_name)
|
7 |
|
8 |
+
tokenizer.src_lang = "nr"
|
9 |
|
10 |
+
model.config.forced_bos_token_id = tokenizer.get_lang_id("en")
|
11 |
+
|
12 |
+
translater_nr_en = pipeline(
|
13 |
+
"translation",
|
14 |
+
model=model,
|
15 |
+
tokenizer=tokenizer,
|
16 |
+
)
|
17 |
|
18 |
def translate(inp):
|
|
|
19 |
res = translater_nr_en(inp, max_length=512, early_stopping=True)[0]['translation_text']
|
20 |
return res
|
21 |
|
22 |
+
# Update description and article as needed
|
23 |
description = """
|
24 |
<p>
|
25 |
<center>
|
|
|
45 |
)
|
46 |
|
47 |
iface.launch(enable_queue=True)
|
48 |
+
|