Roll back to before we wasted our money with Zero

app.py

@@ -10,9 +10,6 @@ import os
 from PIL import Image
 import spaces
 
-print(f"Is CUDA available: {torch.cuda.is_available()}")
-print(f"CUDA device: {torch.cuda.get_device_name(torch.cuda.current_device())}")
-
 from diffusers import (
     StableDiffusionPipeline,
     StableDiffusionControlNetImg2ImgPipeline,
@@ -41,7 +38,7 @@ pipe = StableDiffusionControlNetImg2ImgPipeline.from_pretrained(
     safety_checker=None,
     torch_dtype=torch.float16,
 ).to("cuda")
-# pipe.enable_xformers_memory_efficient_attention()
+pipe.enable_xformers_memory_efficient_attention()
 # pipe.controlnet = torch.nn.DataParallel(pipe.controlnet)
 # pipe.unet = torch.nn.DataParallel(pipe.unet)
 
@@ -83,7 +80,7 @@ SAMPLER_MAP = {
     "DEIS": lambda config: DEISMultistepScheduler.from_config(config),
 }
 
-@spaces.GPU(duration=500)
+@spaces.GPU()
 def inference(
     qr_code_content: str,
     prompt: str,
@@ -327,4 +324,3 @@ model: https://huggingface.co/DionTimmer/controlnet_qrcode-control_v1p_sd15
 os.environ["PYTORCH_CUDA_ALLOC_CONF"] = "expandable_segments:True"
 blocks.queue(max_size=20,api_open=False)
 blocks.launch(share=bool(os.environ.get("SHARE", False)), show_api=False)
-