Facebook researchers propose ‘pre-fine-tuning’ to improve language model performance

Researchers at Facebook propose a new approach to training language models that they claim is highly data efficient.Read More