try clearing memory

app.py

@@ -9,6 +9,7 @@ from sdxl_models import SDXLUNet, SDXLVae, SDXLControlNetPreEncodedControlnetCon
 import torchvision.transforms.functional as TF
 from diffusion import make_sigmas
 from huggingface_hub import hf_hub_download
+import gc
 
 pipe = AutoPipelineForInpainting.from_pretrained("diffusers/stable-diffusion-xl-1.0-inpainting-0.1", torch_dtype=torch.float16, variant="fp16")
 pipe.text_encoder.to("cuda")
@@ -20,6 +21,9 @@ comparing_vae.to(torch.float16)
 comparing_controlnet = SDXLControlNetPreEncodedControlnetCond.load(hf_hub_download("williamberman/sdxl_controlnet_inpainting", "sdxl_controlnet_inpaint_pre_encoded_controlnet_cond_checkpoint_200000.safetensors"))
 comparing_controlnet.to(torch.float16)
 
+gc.collect()
+torch.cuda.empty_cache()
+
 def read_content(file_path: str) -> str:
     """read the content of target file
     """
@@ -51,6 +55,9 @@ def predict(dict, prompt="", negative_prompt="", guidance_scale=7.5, steps=20, s
     pipe.vae.to('cpu')
     pipe.unet.to('cpu')
 
+    gc.collect()
+    torch.cuda.empty_cache()
+
     comparing_unet.to('cuda')
     comparing_vae.to('cuda')
     comparing_controlnet.to('cuda')
@@ -75,6 +82,9 @@ def predict(dict, prompt="", negative_prompt="", guidance_scale=7.5, steps=20, s
     comparing_unet.to('cpu')
     comparing_vae.to('cpu')
     comparing_controlnet.to('cpu')
+
+    gc.collect()
+    torch.cuda.empty_cache()
     
     return output.images[0], out[0], gr.update(visible=True)