fix input bugs

app.py

@@ -26,7 +26,7 @@ def inference(raw_image):
     # raw_image = torch.tensor(raw_image)
     image = transform(raw_image).unsqueeze(0).to(device)
     with torch.no_grad():
-          caption = model.generate(image, sample=False, num_beams=1, max_length=60, min_length=5) 
+          caption = model.generate(image, sample=False, num_beams=1, max_length=200, min_length=5) 
     return caption[0]