Update app.py

app.py

@@ -1,11 +1,11 @@
 import streamlit as st
-from transformers import MarianMTModel, MarianTokenizer
+from transformers import M2M100ForConditionalGeneration, M2M100Tokenizer
 
 # App Title and Description
 st.title("🌐 Universal Language Translator App")
 st.write("""
-Translate text from any language to any other language using an open-source multilingual model.  
-This app supports a wide range of languages with ease of use.
+Translate text from any language to any other language using the open-source M2M100 multilingual model.  
+This app supports over 100 languages and provides a seamless translation experience.
 """)
 
 # Instructions
@@ -41,7 +41,6 @@ languages = {
     "Vietnamese": "vi",
     "Hebrew": "he",
     "Swahili": "sw",
-    "Amharic": "am",
     "Tamil": "ta",
     "Telugu": "te",
     "Punjabi": "pa",
@@ -58,15 +57,18 @@ if st.button("Translate"):
         st.error("Please enter some text to translate.")
     else:
         try:
-            # Universal Model for Language Translation
-            model_name = f"Helsinki-NLP/opus-mt-mul-mul"
-            tokenizer = MarianTokenizer.from_pretrained(model_name)
-            model = MarianMTModel.from_pretrained(model_name)
+            # Load M2M100 model and tokenizer
+            model_name = "facebook/m2m100_418M"
+            tokenizer = M2M100Tokenizer.from_pretrained(model_name)
+            model = M2M100ForConditionalGeneration.from_pretrained(model_name)
             
-            # Translation
-            inputs = tokenizer(f">>{languages[target_language]}<< {source_text}", return_tensors="pt", padding=True, truncation=True)
-            translated_tokens = model.generate(**inputs)
-            translated_text = tokenizer.decode(translated_tokens[0], skip_special_tokens=True)
+            # Set source and target language
+            tokenizer.src_lang = languages[source_language]
+            encoded_text = tokenizer(source_text, return_tensors="pt")
+            generated_tokens = model.generate(**encoded_text, forced_bos_token_id=tokenizer.get_lang_id(languages[target_language]))
+            
+            # Decode the translated text
+            translated_text = tokenizer.decode(generated_tokens[0], skip_special_tokens=True)
             
             # Display Translated Text
             st.subheader("🔄 Translated Text:")