bugfix

app.py

@@ -31,7 +31,7 @@ os.environ['PYTORCH_CUDA_ALLOC_CONF'] = 'max_split_size_mb:30'
 login(token=HF_TOKEN)
 
 MAX_SEED = np.iinfo(np.int32).max
-IMAGE_SIZE = 1024
+IMAGE_SIZE = 864
 
 # init
 device = "cuda" if torch.cuda.is_available() else "cpu"
@@ -179,6 +179,7 @@ def run_flux(
     pipe.enable_model_cpu_offload() # for saving memory
 
     with calculateDuration("run pipe"):
+        print("start to run pipe")
         generated_image = pipe(
             prompt=prompt,
             image=image,
@@ -193,7 +194,7 @@ def run_flux(
             max_sequence_length=256,
             joint_attention_kwargs={"scale": lora_scale}
         ).images[0]
-    clear_cuda_cache()
+    
     return generated_image
 
 @spaces.GPU(duration=120)
@@ -218,7 +219,7 @@ def process(
     bucket:str,
     progress=gr.Progress(track_tqdm=True)
 ):
-    clear_cuda_cache()
+    
     result = {"status": "false", "message": ""}
     if not image_url:
         gr.Info("please enter image url for inpaiting")
@@ -294,7 +295,7 @@ def process(
     
     control_image = control_image.resize((width, height), Image.LANCZOS)    
     control_mode_id = control_mode_ids[control_mode]
-    clear_cuda_cache()
+    
     try:
         generated_image = run_flux(
             image=image,