{"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/multi-modal-open-domain-dialogue","title":"Multi-Modal Open-Domain Dialogue","arxiv_id":"2010.01082","date":"2020-10-02","proceeding":"EMNLP 2021 11","authors":["Kurt Shuster","Eric Michael Smith","Da Ju","Jason Weston"],"abstract":"Recent work in open-domain conversational agents has demonstrated that significant improvements in model engagingness and humanness metrics can be achieved via massive scaling in both pre-training data and model size (Adiwardana et al., 2020; Roller et al., 2020). However, if we want to build agents with human-like abilities, we must expand beyond handling just text. A particularly important topic is the ability to see images and communicate about what is perceived. With the goal of engaging humans in multi-modal dialogue, we investigate combining components from state-of-the-art open-domain dialogue agents with those from state-of-the-art vision models. We study incorporating different image fusion schemes and domain-adaptive pre-training and fine-tuning strategies, and show that our best resulting model outperforms strong existing models in multi-modal dialogue while simultaneously performing as well as its predecessor (text-only) BlenderBot (Roller et al., 2020) in text-based conversation. We additionally investigate and incorporate safety components in our final model, and show that such efforts do not diminish model performance with respect to engagingness metrics.","url_abs":"https://arxiv.org/abs/2010.01082v1","url_pdf":"https://arxiv.org/pdf/2010.01082v1.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":[],"tasks":[{"task_slug":"visual-dialogue","task_name":"Visual Dialog"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/visual-dialog-on-blendedskilltalk","task":"Visual Dialog","dataset":"BlendedSkillTalk","model":"Multi-Modal BlenderBot","rank_in_archive_order":1,"of":1,"metrics":{"BLEU-4":"1","F1":"17.8","ROUGE-L":"19.3"},"uses_additional_data":false},{"leaderboard":"/sota/visual-dialog-on-convai2","task":"Visual Dialog","dataset":"ConvAI2","model":"Multi-Modal BlenderBot","rank_in_archive_order":1,"of":1,"metrics":{"BLEU-4":"1.1","F1":"18.4","ROUGE-L":"22.6"},"uses_additional_data":false},{"leaderboard":"/sota/visual-dialog-on-empatheticdialogues","task":"Visual Dialog","dataset":"EmpatheticDialogues","model":"Multi-Modal BlenderBot","rank_in_archive_order":1,"of":1,"metrics":{"BLEU-4":"1.5","F1":"19.2","ROUGE-L":"24.5"},"uses_additional_data":false},{"leaderboard":"/sota/visual-dialog-on-image-chat","task":"Visual Dialog","dataset":"Image-Chat","model":"Multi-Modal BlenderBot","rank_in_archive_order":1,"of":1,"metrics":{"BLEU-4":"40","F1":"13.1","ROUGE-L":"18"},"uses_additional_data":false},{"leaderboard":"/sota/visual-dialog-on-wizard-of-wikipedia","task":"Visual Dialog","dataset":"Wizard of Wikipedia","model":"Multi-Modal BlenderBot","rank_in_archive_order":1,"of":1,"metrics":{"BLEU-4":"2.2","F1":"18.6","ROUGE-L":"17.4"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2010.01082","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}