Update app.py

app.py

@@ -65,11 +65,18 @@ ddg = DuckDuckGoSearchTool()
 # custom_role_conversions=None,
 # )
 
-model = LiteLLMModel(
-    model_id="ollama_chat/qwen2.5:7b",        # Ollama model identifier
-    api_base="http://127.0.0.1:11434",        # default Ollama HTTP port
-    num_ctx=8192,                            
-    temperature=0.5                           
+# model = LiteLLMModel(
+#     model_id="ollama_chat/qwen2.5:7b",        # Ollama model identifier
+#     api_base="http://127.0.0.1:11434",        # default Ollama HTTP port
+#     num_ctx=8192,                            
+#     temperature=0.5                           
+# )
+
+model = LocalModel(
+    model_id="TheBloke/vicuna-7B-1.1-HF",
+    trust_remote_code=True,   # needed if the repo defines its own model classes
+    device="cuda",            # or "cpu" if no GPU
+    quantize="4bit"           # optional: drastically reduce VRAM usage
 )
 
 # Import tool from Hub