add HF_TOKEN_INFERENCE_API

app.py

@@ -1,3 +1,4 @@
+import os
 import json
 import time
 from itertools import count, islice
@@ -17,7 +18,7 @@ from utils import StringIteratorIO
 
 
 model_id = "meta-llama/Meta-Llama-3.1-8B-Instruct"
-client = InferenceClient(model_id)
+client = InferenceClient(model_id, token=os.environ.get("HF_TOKEN_INFERENCE_API"))
 
 session = requests.Session()
 empty_dataframe = pd.DataFrame({"1": [], "2": [], "3": []})