refactor: update InferenceClient configuration and specify llama model in chat completion request

app.py

@@ -6,8 +6,8 @@ import os
 app = FastAPI()
 
 client = InferenceClient(
-    model="nvidia/Llama-3.1-Nemotron-Nano-8B-v1",
-    token=os.environ.get("HF_TOKEN")
+    provider="hf-inference",
+    api_key=os.environ.get("HF_TOKEN")
 )
 
 class ChatRequest(BaseModel):
@@ -20,6 +20,7 @@ def greet_json():
 @app.post("/chat")
 def chat(request: ChatRequest):
     response = client.chat.completions.create(
+        model="meta-llama/Llama-3.1-8B-Instruct",
         messages=[{"role": "user", "content": request.message}],
         max_tokens=1024,
     )