Update app.py

app.py

@@ -3,8 +3,8 @@ import gradio as gr
 from huggingface_hub import InferenceClient
 
 HF_TOKEN = os.environ.get("HF_TOKEN", None)
-API_URL = "https://api-inference.huggingface.co/models/meta-llama/Llama-2-70b-chat-hf"
-API_URL_2 = "https://api-inference.huggingface.co/models/codellama/CodeLlama-34b-Instruct-hf"
+API_URL = "meta-llama/Llama-2-70b-chat-hf"
+API_URL_2 = "codellama/CodeLlama-34b-Instruct-hf"
 BOT_NAME = "Assistant"
 
 STOP_SEQUENCES = ["\nUser:", " User:", "###", "</s>"]
@@ -19,12 +19,12 @@ EXAMPLES = [
 
 client = InferenceClient(
     API_URL,
-    headers={"Authorization": f"Bearer {HF_TOKEN}"},
+    token=HF_TOKEN,
 )
 
 client2 = InferenceClient(
     API_URL_2,
-    headers={"Authorization": f"Bearer {HF_TOKEN}"},
+    token=HF_TOKEN,
 )
 
 def format_prompt(message, history, system_prompt):