Published onSeptember 19, 2019Fine-tuning GPT-2 from human preferencesGPT-2openairesearchGPT-2, fine-tuned the 774M parameter GPT-2 language model using human feedback for various tasks.