use lru_cache to cache performance for already seen prompts

main.py

-from functools import partial
+from functools import lru_cache, partial
 from pathlib import Path
 from typing import DefaultDict
 
@@ -169,7 +169,7 @@ if __name__ == "__main__":
     D = load_dataset("SetFit/sst2", split="validation[:200]")
 
     # - fD(·) denotes the score of a prompt on the desired LLM evaluated on D
-    f_D = partial(evaluate_prompt, dataset=D)
+    f_D = lru_cache(maxsize=None)(partial(evaluate_prompt, dataset=D))
 
     # - a pre-defined number of iterations T
     T = 10