Update app.py

app.py

@@ -1,4 +1,4 @@
-from smolagents import CodeAgent, DuckDuckGoSearchTool, HfApiModel, load_tool, tool, LiteLLMModel
+from smolagents import CodeAgent, DuckDuckGoSearchTool, HfApiModel, load_tool, tool, LiteLLMModel, LocalModel
 import datetime
 import requests
 import pytz
@@ -65,15 +65,19 @@ 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 = OllamaQwen7BClientModel()
+# 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",  # or another HF repo you prefer
+    trust_remote_code=True,               # allow custom model code
+    device="cuda",                        # or "cpu" if no GPU
+    quantize="4bit"                       # drastically reduce VRAM use
+)
 
 # Import tool from Hub
 image_generation_tool = load_tool("agents-course/text-to-image", trust_remote_code=True)