Update app.py

app.py

@@ -11,8 +11,8 @@ os.environ['PYTORCH_CUDA_ALLOC_CONF'] = 'max_split_size_mb:54'
 device = "cuda" if torch.cuda.is_available() else "cpu"
 
 tokenizer = YiTokenizer(vocab_file="./tokenizer.model")
-model = AutoModelForCausalLM.from_pretrained(model_id, load_in_8bit=True, trust_remote_code=True)
-# model = model.to(torch.bfloat16)
+model = AutoModelForCausalLM.from_pretrained(model_id, trust_remote_code=True)
+model = model.to(torch.bfloat16)
 model = model.to(device)
 
 def run(message, chat_history, max_new_tokens=4056, temperature=3.5, top_p=0.9, top_k=800):