Update app.py

app.py

@@ -45,11 +45,13 @@ def load_model():
         # Load processor
         processor = AutoProcessor.from_pretrained(base_model_id)
         
-        # Load the pre-quantized model
+        # Load the pre-quantized model with unsloth settings
         model = AutoModelForCausalLM.from_pretrained(
             base_model_id,
             device_map="auto",
-            torch_dtype=torch.float16
+            torch_dtype=torch.float16,
+            use_cache=True,
+            trust_remote_code=True
         )
         
         # Load adapter