To build a high-quality open-domain chatbot, we introduce the effective training process of PLATO-2 via curriculum learning. There are two stages involved in the learning process. In the first stage, a coarse-grained generation model is trained to learn response generation under the simplified framework of one-to-one mapping. In the second stage, a fine-grained generative model augmented with latent variables and an evaluation model are further trained to generate diverse responses and to select the best response, respectively. PLATO-2 was trained on both Chinese and English data, whose effectiveness and superiority are verified through comprehensive evaluations, achieving new state-of-the-art results.
@article{arxiv.2006.16779,
title = {PLATO-2: Towards Building an Open-Domain Chatbot via Curriculum Learning},
author = {Siqi Bao and Huang He and Fan Wang and Hua Wu and Haifeng Wang and Wenquan Wu and Zhen Guo and Zhibin Liu and Xinchao Xu},
journal= {arXiv preprint arXiv:2006.16779},
year = {2021}
}
Comments
Findings of ACL 2021. First four authors contributed equally to this work