| import os
|
| import ollama
|
| from langchain_ollama import OllamaLLM
|
| from langchain_openai import ChatOpenAI
|
| from config import config
|
|
|
| def get_llm(model_name: str = None, temperature: float = 0.7):
|
| """Return LLM (Cloud API if API_KEY is set, otherwise Ollama)"""
|
| if model_name is None:
|
| model_name = config.MAIN_MODEL
|
|
|
| if config.API_KEY:
|
| return ChatOpenAI(
|
| model=model_name,
|
| temperature=temperature,
|
| api_key=config.API_KEY,
|
| base_url=config.BASE_URL
|
| )
|
|
|
|
|
| ollama_base_url = os.getenv("OLLAMA_HOST", "http://localhost:11434")
|
| return OllamaLLM(
|
| model=model_name,
|
| temperature=temperature,
|
| base_url=ollama_base_url
|
| )
|
|
|
| def get_fast_llm(temperature: float = 0.5):
|
| """Fast model for mutation and judging"""
|
| return get_llm(model_name=config.FAST_MODEL, temperature=temperature)
|
|
|
| def call_llm(prompt: str, model: str = None, temperature: float = 0.7) -> str:
|
| """Universal LLM call: Uses Cloud API if key exists, otherwise Ollama"""
|
| if config.API_KEY:
|
| llm = ChatOpenAI(
|
| model=model or config.MAIN_MODEL,
|
| temperature=temperature,
|
| api_key=config.API_KEY,
|
| base_url=config.BASE_URL
|
| )
|
| return llm.invoke(prompt).content
|
|
|
|
|
| if model is None:
|
| model = config.MAIN_MODEL
|
| response = ollama.chat(model=model, messages=[{'role': 'user', 'content': prompt}])
|
| return response['message']['content']
|
|
|
| def save_best_prompt(name: str, prompt: str, score: float):
|
| """Save successful evolved prompts"""
|
| import json
|
| from datetime import datetime
|
| import os
|
|
|
| data = {
|
| "name": name,
|
| "prompt": prompt,
|
| "score": score,
|
| "date": datetime.now().strftime("%Y-%m-%d %H:%M")
|
| }
|
|
|
| filepath = os.path.join(config.BEST_PROMPTS_DIR, f"{name.replace(' ', '_')}.json")
|
| with open(filepath, "w", encoding="utf-8") as f:
|
| json.dump(data, f, indent=2)
|
|
|
| print(f"[DISK] Saved best prompt: {name} (Score: {score:.2f})") |
