Refactor generate_llm function to remove max_tokens and temperature parameters; use hardcoded defaults instead.

polLLM.py

@@ -22,7 +22,6 @@ logger.addHandler(handler)
 _DEFAULT_MODEL     = "openai-large" #_config.get("model", "openai-large")
 _SYSTEM_TEMPLATE   = _config.get("system_prompt", "")
 _CHAR              = _config.get("char", "Eve")
-_DEFAULT_MAX_TOKENS = 8000 #_config.get("max_tokens", 8000)
 
 # --- OpenAI client init ---
 client = OpenAI(
@@ -39,17 +38,12 @@ def _build_system_prompt() -> str:
 def generate_llm(
     prompt: str,
     model: str = None,
-    max_tokens: int = None,
-    temperature: float = None,
 ) -> str:
     """
     Send a chat-completion request to the LLM, with retries and backoff.
     Reads defaults from config.yaml, but can be overridden per-call.
     """
     model       = model or _DEFAULT_MODEL
-    max_tokens  = max_tokens or _DEFAULT_MAX_TOKENS
-    temperature = temperature if temperature is not None else _DEFAULT_TEMPERATURE
-
     system_prompt = _build_system_prompt()
     messages = [
         {"role": "system", "content": system_prompt},
@@ -64,8 +58,6 @@ def generate_llm(
             resp = client.chat.completions.create(
                 model       = model,
                 messages    = messages,
-                max_tokens  = max_tokens,
-                temperature = temperature,
                 seed        = seed,
             )
             text = resp.choices[0].message.content.strip()