Update app.py

app.py

@@ -7,10 +7,18 @@ import os
 MODEL_NAME = "model.safetensors"  # Replace with your actual HF model repo
 DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
 
-# Load model and tokenizer
-@gr.utils.cache
+# Global variables for model caching
+_model = None
+_tokenizer = None
+
 def load_model():
-    """Load the model and tokenizer with caching"""
+    """Load the model and tokenizer with simple caching"""
+    global _model, _tokenizer
+    
+    # Return cached model if already loaded
+    if _model is not None and _tokenizer is not None:
+        return _model, _tokenizer
+    
     print(f"Loading model from: {MODEL_NAME}")
     print(f"Using device: {DEVICE}")
     
@@ -33,6 +41,11 @@ def load_model():
         model = model.to(DEVICE)
     
     print("✅ Model and tokenizer loaded successfully!")
+    
+    # Cache the loaded model and tokenizer
+    _model = model
+    _tokenizer = tokenizer
+    
     return model, tokenizer
 
 # Initialize model and tokenizer