Update app_chat.py

app_chat.py

@@ -58,7 +58,7 @@ def generate(
     conversation += chat_history
     conversation.append({"role": "user", "content": message})
 
-    input_ids = tokenizer.apply_chat_template(conversation,  tokenize=True, add_generation_prompt=True, return_tensors="pt").to('cuda').to(torch.bfloat16)
+    input_ids = tokenizer.apply_chat_template(conversation,  tokenize=True, add_generation_prompt=True, return_tensors="pt").to('cuda')
     
     # stopping_criteria = StoppingCriteriaList([StopStringCriteria(tokenizer=tokenizer, stop_strings="</s>")])