الفصل 3: إدارة التكلفة

استراتيجيات التخزين المؤقت

import hashlib
from functools import lru_cache
import redis

class LLMCache:
    def __init__(self):
        self.redis = redis.Redis()
        self.ttl = 3600  # 1 hour
    
    def _hash_request(self, prompt: str, params: dict) -> str:
        key = f"{prompt}:{json.dumps(params, sort_keys=True)}"
        return hashlib.sha256(key.encode()).hexdigest()
    
    def get(self, prompt: str, params: dict) -> str | None:
        key = self._hash_request(prompt, params)
        cached = self.redis.get(key)
        return cached.decode() if cached else None
    
    def set(self, prompt: str, params: dict, response: str):
        key = self._hash_request(prompt, params)
        self.redis.setex(key, self.ttl, response)

# Usage
cache = LLMCache()

def generate_with_cache(prompt: str, params: dict):
    cached = cache.get(prompt, params)
    if cached:
        return cached  # Cache hit - free!
    
    response = llm.generate(prompt, **params)
    cache.set(prompt, params, response)
    return response

اختيار النموذج بناءً على التكلفة

class CostAwareRouter:
    def __init__(self):
        self.models = {
            "simple": {"model": "gpt-3.5-turbo", "cost_per_1k": 0.002},
            "complex": {"model": "gpt-4", "cost_per_1k": 0.06}
        }
    
    def classify_complexity(self, prompt: str) -> str:
        # Simple heuristics
        if len(prompt) < 100 and "?" in prompt:
            return "simple"
        return "complex"
    
    def route(self, prompt: str) -> str:
        complexity = self.classify_complexity(prompt)
        return self.models[complexity]["model"]

---

التالي: اختبار الوحدة 11