Is OpenAI's Reinforcement Fine-Tuning (RFT) Worth It?