Open DryPilgrim opened 8 months ago
//cn 1.在data collection中没有提到train set,它是怎么构造的?数据有开源吗? 2.data collection中提到的development set和test set都是用来评估模型的吗? 3.test set总共只有157条数据,如何确保测试结果的可靠性? //en
@DryPilgrim From my understanding, this is strictly prompt based approach, there is no need of a fine-tuning, therefore a train set does not exist
//cn 1.在data collection中没有提到train set,它是怎么构造的?数据有开源吗? 2.data collection中提到的development set和test set都是用来评估模型的吗? 3.test set总共只有157条数据,如何确保测试结果的可靠性? //en