Update app.py

app.py

@@ -5,7 +5,6 @@ import torch
 import os, gc
 import httpx
 
-
 device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
 # Language options and mappings
 favourite_langs = {"Romanian": "ro", "German": "de", "English": "en", "-----": "-----"}
@@ -146,6 +145,7 @@ class Translators:
             prompt = f"Translate the following segment into {self.tl}, without additional explanation.\n\n{self.input_text}."
         tokenizer = AutoTokenizer.from_pretrained(self.model_name)
         model = AutoModelForCausalLM.from_pretrained(self.model_name, device_map="auto")
+        model.tie_weights() # fp8
         messages = [{"role": "user", "content": prompt}]
         tokenized_chat = tokenizer.apply_chat_template(
             messages,