Update app.py

app.py

@@ -1,4 +1,3 @@
-from huggingface_hub import InferenceClient
 from fastapi import FastAPI, Request
 from pydantic import BaseModel
 import openai
@@ -14,8 +13,6 @@ HF_TOKEN = os.environ["HF_TOKEN"]
 HB_TOKEN = os.environ["HB_TOKEN"]
 PROMPTS_DOC_URL = os.environ["PROMPTS"]
 
-client = InferenceClient(model=MODEL, token=HF_TOKEN)
-
 def fetch_prompts_from_google_doc():
     print("Fetching prompts from Google Doc...")
     response = requests.get(PROMPTS_DOC_URL)
@@ -66,6 +63,11 @@ async def drive(prompt: DrivePrompt):
     {prompt.code}
     """
 
+    client = openai.OpenAI(
+        api_key=HB_TOKEN,
+        base_url="https://api.hyperbolic.xyz/v1",
+    )
+
     chat_completion = client.chat.completions.create(
         model="meta-llama/Meta-Llama-3-70B-Instruct",
         messages=[
@@ -115,8 +117,8 @@ async def chat(prompt: NavigatePrompt):
     """
     
     client = openai.OpenAI(
-    api_key=HB_TOKEN,
-    base_url="https://api.hyperbolic.xyz/v1",
+        api_key=HB_TOKEN,
+        base_url="https://api.hyperbolic.xyz/v1",
     )
 
     chat_completion = client.chat.completions.create(