Update app.py

app.py

@@ -61,7 +61,7 @@ accelerator = Accelerator()
 model = AutoModelForCausalLM.from_pretrained(model_id, token= token, 
                                                  # torch_dtype= torch.uint8, 
                                              torch_dtype=torch.bfloat16,
-                                              load_in_8bit=True,
+                                              load_in_4bit=True,
                                              # #  # torch_dtype=torch.fl,
                                              attn_implementation="flash_attention_2",
                                              low_cpu_mem_usage=True,