Update app.py

app.py

@@ -15,7 +15,7 @@ MODELS = {
     'Bengali': "Helsinki-NLP/opus-mt-en-bn",
 }
 
-# Manually defined language codes for different language models
+# Language codes for different models
 LANG_CODE_MAP = {
     'French': 'fr',
     'Spanish': 'es',
@@ -38,16 +38,16 @@ def translate_text(text, target_lang):
     model = MarianMTModel.from_pretrained(model_name)
     tokenizer = MarianTokenizer.from_pretrained(model_name)
     
-    # Encode the text and prepare it for translation
+    # Encode the text
     encoded_text = tokenizer(text, return_tensors="pt")
     
-    # Get the language code for forced_bos_token_id
+    # Language code for forced_bos_token_id
     lang_code = LANG_CODE_MAP.get(target_lang)
     if not lang_code:
         return "Error: Language code not found."
     
     # Translate text
-    translated = model.generate(**encoded_text, forced_bos_token_id=tokenizer.lang_code_to_id.get(lang_code))
+    translated = model.generate(**encoded_text, forced_bos_token_id=tokenizer.convert_tokens_to_ids(lang_code))
     
     # Decode the translated text
     translated_text = tokenizer.decode(translated[0], skip_special_tokens=True)