Add CUDA cache freeing before model re-loading

app.py

@@ -1,6 +1,8 @@
 import os
 import sys
 from pathlib import Path
+import gc
+import torch
 import gradio as gr
 
 # Allow importing your models package
@@ -31,6 +33,14 @@ def load_model_with_quantization(quantization_mode: str):
     """Load or reload the model with specified quantization"""
     global model, current_quantization
     
+    # Free GPU memory if model already exists
+    if model is not None:
+        print("Unloading existing model and freeing GPU memory...")
+        del model
+        gc.collect()
+        torch.cuda.empty_cache()
+        print("GPU memory cleared.")
+    
     load_8bit = False
     load_4bit = False