Update app.py

app.py

@@ -98,7 +98,7 @@ pipe.to(device=device, dtype=torch.bfloat16)
 
 tokenizer = AutoTokenizer.from_pretrained(checkpoint, add_prefix_space=False)
 tokenizer.tokenizer_legacy=False
-model = AutoModelForCausalLM.from_pretrained(checkpoint,attn_implementation="flash_attention_2").to('cuda')
+model = AutoModelForCausalLM.from_pretrained(checkpoint).to('cuda')
 #model = torch.compile(model)
 
 upscaler_2 = UpscaleWithModel.from_pretrained("Kim2091/ClearRealityV1") #.to(torch.device("cuda:0"))