Openai Fine Tuning Training Loss