Running LLM Long Output Experiment (Code Generation) 📈 Evaluating max single output length of code gen LLMs