import streamlit as st from transformers import MarianMTModel, MarianTokenizer # Define a dictionary to map language names to model identifiers models = { 'French': 'Helsinki-NLP/opus-mt-en-fr', 'Spanish': 'Helsinki-NLP/opus-mt-en-es', 'German': 'Helsinki-NLP/opus-mt-en-de', 'Italian': 'Helsinki-NLP/opus-mt-en-it', 'Urdu': 'Helsinki-NLP/opus-mt-en-ur', 'Arabic': 'Helsinki-NLP/opus-mt-en-ar', # Add more language models if needed } def load_model(model_name): """Load the model and tokenizer based on the selected model name.""" model = MarianMTModel.from_pretrained(model_name) tokenizer = MarianTokenizer.from_pretrained(model_name) return model, tokenizer def translate_text(text, model, tokenizer): """Translate text using the provided model and tokenizer.""" inputs = tokenizer.encode(text, return_tensors="pt") translated = model.generate(inputs) translated_text = tokenizer.decode(translated[0], skip_special_tokens=True) return translated_text def main(): st.title("Multilingual Translator") # User input for English text text_to_translate = st.text_area("Enter text in English:") # Language options selected_language = st.selectbox("Select target language:", list(models.keys())) if st.button("Translate"): if text_to_translate: # Load the selected model model_name = models[selected_language] model, tokenizer = load_model(model_name) translated_text = translate_text(text_to_translate, model, tokenizer) st.write(f"**Translation in {selected_language}:**") st.write(translated_text) else: st.warning("Please enter text to translate.") if __name__ == "__main__": main() try: import transformers from transformers import MarianMTModel, MarianTokenizer print("Transformers imported successfully") except ImportError as e: print(f"ImportError: {e}") python test_import.py