Model Change

Switched to Qwen3-4B-Instruct-2507, a better fitting model for the anticipated task types.

app.py

@@ -18,13 +18,13 @@ if "HUGGINGFACEHUB_API_TOKEN" not in os.environ:
 
 # --- LLM and Template Configuration ---
 llm = HuggingFaceEndpoint(
-    repo_id="HuggingFaceH4/zephyr-7b-alpha",  # inference-ready model
+    repo_id="Qwen/Qwen3-4B-Instruct-2507",
     temperature=0.7,
-    max_new_tokens=512,
-    huggingfacehub_api_token=os.getenv("HUGGINGFACEHUB_API_TOKEN"),
-    task="conversational"   
+    top_p=0.8,
+    top_k=20,
+    max_new_tokens=1024,
+    huggingfacehub_api_token=os.getenv("HUGGINGFACEHUB_API_TOKEN")
 )
-
 math_template = ChatPromptTemplate.from_messages([
     ("system", """{system_message}
 You are an expert math tutor. For every math problem: