added use of ZERO so it runs on GPU

app.py

@@ -1,16 +1,20 @@
 import gradio as gr
 from PIL import Image
+import spaces
 import torch
 from transformers import BlipProcessor, BlipForConditionalGeneration
 
-device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
 
+device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
 processor = BlipProcessor.from_pretrained("noamrot/FuseCap_Image_Captioning")
 model = BlipForConditionalGeneration.from_pretrained("noamrot/FuseCap_Image_Captioning").to(device)
 
+@spaces.GPU(duration=15)
 def inference(raw_image):
+    cur_device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
+    model = model.to(device)
     text = "a picture of "
-    inputs = processor(raw_image, text, return_tensors="pt").to(device)
+    inputs = processor(raw_image, text, return_tensors="pt").to(cur_device)
     out = model.generate(**inputs)
     caption = processor.decode(out[0], skip_special_tokens=True)
     return caption