{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/galaxy-a-generative-pre-trained-model-for","title":"GALAXY: A Generative Pre-trained Model for Task-Oriented Dialog with Semi-Supervised Learning and Explicit Policy Injection","arxiv_id":"2111.14592","date":"2021-11-29","proceeding":null,"authors":["Wanwei He","Yinpei Dai","Yinhe Zheng","Yuchuan Wu","Zheng Cao","Dermot Liu","Peng Jiang","Min Yang","Fei Huang","Luo Si","Jian Sun","Yongbin Li"],"abstract":"Pre-trained models have proved to be powerful in enhancing task-oriented dialog systems. However, current pre-training methods mainly focus on enhancing dialog understanding and generation tasks while neglecting the exploitation of dialog policy. In this paper, we propose GALAXY, a novel pre-trained dialog model that explicitly learns dialog policy from limited labeled dialogs and large-scale unlabeled dialog corpora via semi-supervised learning. Specifically, we introduce a dialog act prediction task for policy optimization during pre-training and employ a consistency regularization term to refine the learned representation with the help of unlabeled dialogs. We also implement a gating mechanism to weigh suitable unlabeled dialog samples. Empirical results show that GALAXY substantially improves the performance of task-oriented dialog systems, and achieves new state-of-the-art results on benchmark datasets: In-Car, MultiWOZ2.0 and MultiWOZ2.1, improving their end-to-end combined scores by 2.5, 5.3 and 5.5 points, respectively. We also show that GALAXY has a stronger few-shot ability than existing models under various low-resource settings.","url_abs":"https://arxiv.org/abs/2111.14592v8","url_pdf":"https://arxiv.org/pdf/2111.14592v8.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"galaxy-a-generative-pre-trained-model-for","repo_url":"https://github.com/siat-nlp/galaxy","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"end-to-end-dialogue-modelling","task_name":"End-To-End Dialogue Modelling"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/end-to-end-dialogue-modelling-on-multiwoz-2-0","task":"End-To-End Dialogue Modelling","dataset":"MULTIWOZ 2.0","model":"GALAXY","rank_in_archive_order":1,"of":6,"metrics":{"BLEU":"20.5","MultiWOZ (Inform)":"94.4","MultiWOZ (Success)":"85.3"},"uses_additional_data":false},{"leaderboard":"/sota/end-to-end-dialogue-modelling-on-multiwoz-2-1","task":"End-To-End Dialogue Modelling","dataset":"MULTIWOZ 2.1","model":"GALAXY","rank_in_archive_order":1,"of":4,"metrics":{"BLEU":"20.01","MultiWOZ (Inform)":"95.30","MultiWOZ (Success)":"86.20"},"uses_additional_data":false}],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=2111.14592","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}