added model download

app.py

@@ -20,9 +20,9 @@ def download_model():
 
 def load_model():
     from llama_cpp import Llama, LlamaGrammar
-    model_url="https://huggingface.co/TheBloke/Llama-2-7B-GGUF/resolve/main/llama-2-7b.Q5_K_S.gguf"
+    model_file="llama-2-7b.Q5_K_S.gguf"
     llm = Llama(
-        model_path=model_url,
+        model_path=model_file,
         n_gpu_layers=-1, verbose=False
     )