On LLM-generated Logic Programs and their Inference Execution Methods