Update app.py

app.py

@@ -61,9 +61,9 @@ vqllm = AutoModelForCausalLM.from_pretrained(
     model_id,
     attn_implementation='flash_attention_2',
     torch_dtype=torch.bfloat16,
-    load_in_8bit=True,
-    max_memory={0: "40GiB" },
-    ) # .to("cuda:0")
+    # load_in_8bit=True,
+    # max_memory={0: "40GiB" },
+    ).to("cuda:0")
 
 stop_flag = False