LLM shall not be called if seed is frozen and cache is activated.
This allows to reproduce behavior and have reproductible python applications.
It also answer to gthe comment from INRIA report 1-a:
Cette approche
permet de simuler rapidement des comportements sans implémentation explicite, mais présente
l’inconvénient de faire appel à un LLM lors de chaque exécution même pour des traitements simples.
LLM shall not be called if seed is frozen and cache is activated.
This allows to reproduce behavior and have reproductible python applications.
It also answer to gthe comment from INRIA report 1-a: