fix: HF_TOKEN variable, max_steps from task, document data in prompt

inference.py

@@ -28,7 +28,7 @@ from env import InvoiceExceptionEnv, Action, ALL_TASKS
 
 API_BASE_URL = os.getenv("API_BASE_URL", "https://router.huggingface.co/v1")
 MODEL_NAME = os.getenv("MODEL_NAME", "Qwen/Qwen2.5-72B-Instruct")
-API_KEY = os.getenv("HF_TOKEN") or os.getenv("API_KEY", "")
+HF_TOKEN = os.getenv("HF_TOKEN")
 
 
 # ---------------------------------------------------------------------------
@@ -294,7 +294,7 @@ def run_task(client: OpenAI, env: InvoiceExceptionEnv, task_id: str) -> tuple:
 
 def main() -> None:
     """Run inference on all tasks."""
-    client = OpenAI(base_url=API_BASE_URL, api_key=API_KEY)
+    client = OpenAI(base_url=API_BASE_URL, api_key=HF_TOKEN)
     env = InvoiceExceptionEnv(seed=42)
 
     all_scores = []