Upload inference.py

inference.py

@@ -6,11 +6,12 @@ from models import RiderSafetyAction, RiderSafetyObservation
 from client import RiderSafetyClient 
 
 # 1. Configuration
-API_BASE_URL = os.getenv("API_BASE_URL", "https://api-inference.huggingface.co/v1/")
-MODEL_NAME = os.getenv("MODEL_NAME", "meta-llama/Meta-Llama-3-8B-Instruct")
-HF_TOKEN = os.getenv("HF_TOKEN")
+# The validator requires using API_BASE_URL and API_KEY environment variables exactly.
+API_BASE_URL = os.environ.get("API_BASE_URL")
+API_KEY = os.environ.get("API_KEY")
+MODEL_NAME = os.environ.get("MODEL_NAME", "meta-llama/Meta-Llama-3-8B-Instruct")
 
-client = OpenAI(base_url=API_BASE_URL, api_key=HF_TOKEN)
+client = OpenAI(base_url=API_BASE_URL, api_key=API_KEY)
 
 # 2. Logging Helpers (Standard Format)
 def log_start(task, env_name, model):
@@ -93,7 +94,7 @@ async def run_task(env: RiderSafetyClient, task_name: str):
         log_end(success_flag, steps, score, rewards)
 
 async def main():
-    env = RiderSafetyClient("http://localhost:8000")
+    env = RiderSafetyClient("http://localhost:7860")
     for task in ["easy", "medium", "hard"]:
         await run_task(env, task)
         print("-" * 50)