About learning, the ERNIE language model has continuous learning meaning it has not the necessity to fully retrain (I believe) Also GPT 3 is obscoleted by orde

About learning, the ERNIE language model has continuous learning meaning it has not the necessity to fully retrain (I believe) Also GPT 3 is obscoleted by order of magnitudes by SMIM https://arxiv.org/abs/2003.02645 https://arxiv.org/abs/2003.02645