Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import streamlit as st
|
| 2 |
+
from transformers import MarianMTModel, MarianTokenizer
|
| 3 |
+
|
| 4 |
+
# Function to load model and tokenizer based on selected languages
|
| 5 |
+
def load_model_and_tokenizer(src_lang, tgt_lang):
|
| 6 |
+
model_name = f"Helsinki-NLP/opus-mt-{src_lang}-{tgt_lang}"
|
| 7 |
+
model = MarianMTModel.from_pretrained(model_name)
|
| 8 |
+
tokenizer = MarianTokenizer.from_pretrained(model_name)
|
| 9 |
+
return model, tokenizer
|
| 10 |
+
|
| 11 |
+
# Function to translate text
|
| 12 |
+
def translate_text(text, model, tokenizer):
|
| 13 |
+
inputs = tokenizer(text, return_tensors="pt", padding=True)
|
| 14 |
+
translated = model.generate(**inputs)
|
| 15 |
+
return tokenizer.decode(translated[0], skip_special_tokens=True)
|
| 16 |
+
|
| 17 |
+
# Streamlit app layout
|
| 18 |
+
st.title("Language Translator App")
|
| 19 |
+
st.write("Translate text between multiple languages using open-source models.")
|
| 20 |
+
|
| 21 |
+
# Language options (expand this list based on available models)
|
| 22 |
+
languages = {
|
| 23 |
+
"English": "en",
|
| 24 |
+
"French": "fr",
|
| 25 |
+
"Spanish": "es",
|
| 26 |
+
"German": "de",
|
| 27 |
+
"Italian": "it",
|
| 28 |
+
"Russian": "ru",
|
| 29 |
+
"Chinese": "zh"
|
| 30 |
+
}
|
| 31 |
+
|
| 32 |
+
# Language selection
|
| 33 |
+
source_language = st.selectbox("Select Input Language:", options=list(languages.keys()))
|
| 34 |
+
target_language = st.selectbox("Select Output Language:", options=list(languages.keys()))
|
| 35 |
+
|
| 36 |
+
# Check if source and target languages are different
|
| 37 |
+
if source_language == target_language:
|
| 38 |
+
st.warning("Input and output languages must be different.")
|
| 39 |
+
else:
|
| 40 |
+
src_lang_code = languages[source_language]
|
| 41 |
+
tgt_lang_code = languages[target_language]
|
| 42 |
+
|
| 43 |
+
# Load model and tokenizer
|
| 44 |
+
with st.spinner("Loading translation model..."):
|
| 45 |
+
model, tokenizer = load_model_and_tokenizer(src_lang_code, tgt_lang_code)
|
| 46 |
+
|
| 47 |
+
# Input text
|
| 48 |
+
text_to_translate = st.text_area("Enter text to translate:")
|
| 49 |
+
|
| 50 |
+
if st.button("Translate"):
|
| 51 |
+
if text_to_translate.strip():
|
| 52 |
+
with st.spinner("Translating..."):
|
| 53 |
+
translated_text = translate_text(text_to_translate, model, tokenizer)
|
| 54 |
+
st.success("Translation Completed!")
|
| 55 |
+
st.text_area("Translated Text:", value=translated_text, height=200)
|
| 56 |
+
else:
|
| 57 |
+
st.error("Please enter text to translate.")
|