Spaces:
Sleeping
Sleeping
Create main.py
Browse files
main.py
ADDED
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
from transformers import MarianMTModel, MarianTokenizer
|
3 |
+
|
4 |
+
# Define available languages with MarianMT models
|
5 |
+
LANGUAGES = {
|
6 |
+
'Spanish': 'es',
|
7 |
+
'French': 'fr',
|
8 |
+
'German': 'de',
|
9 |
+
'Chinese': 'zh',
|
10 |
+
'Hindi': 'hi',
|
11 |
+
'Arabic': 'ar',
|
12 |
+
'Japanese': 'ja',
|
13 |
+
'Russian': 'ru',
|
14 |
+
'Italian': 'it',
|
15 |
+
'Portuguese': 'pt',
|
16 |
+
# Add more languages if needed
|
17 |
+
}
|
18 |
+
|
19 |
+
# Function to load the model based on the selected language
|
20 |
+
@st.cache_resource
|
21 |
+
def load_model(src_lang='en', tgt_lang='es'):
|
22 |
+
model_name = f'Helsinki-NLP/opus-mt-{src_lang}-{tgt_lang}'
|
23 |
+
model = MarianMTModel.from_pretrained(model_name)
|
24 |
+
tokenizer = MarianTokenizer.from_pretrained(model_name)
|
25 |
+
return model, tokenizer
|
26 |
+
|
27 |
+
# Function to translate text
|
28 |
+
def translate_text(model, tokenizer, text):
|
29 |
+
inputs = tokenizer.encode(text, return_tensors='pt', truncation=True, padding=True)
|
30 |
+
translated = model.generate(inputs, max_length=512, num_beams=5, early_stopping=True)
|
31 |
+
translated_text = tokenizer.decode(translated[0], skip_special_tokens=True)
|
32 |
+
return translated_text
|
33 |
+
|
34 |
+
# Streamlit app
|
35 |
+
st.title("Language Translator")
|
36 |
+
st.write("Translate English text to any language using Hugging Face models.")
|
37 |
+
|
38 |
+
# Input text
|
39 |
+
text = st.text_area("Enter text in English to translate:")
|
40 |
+
|
41 |
+
# Language selection
|
42 |
+
language = st.selectbox("Choose target language", list(LANGUAGES.keys()))
|
43 |
+
|
44 |
+
if st.button("Translate"):
|
45 |
+
if text:
|
46 |
+
# Load model and tokenizer based on selected language
|
47 |
+
tgt_lang = LANGUAGES[language]
|
48 |
+
model, tokenizer = load_model('en', tgt_lang)
|
49 |
+
|
50 |
+
# Perform translation
|
51 |
+
translated_text = translate_text(model, tokenizer, text)
|
52 |
+
|
53 |
+
# Display the translation
|
54 |
+
st.write(f"**Translated text ({language}):**")
|
55 |
+
st.write(translated_text)
|
56 |
+
else:
|
57 |
+
st.write("Please enter text to translate.")
|