--- title: "Training Millions of Personalized Dialogue — Mazaré 2018" description: "End-to-end chit-chat models are often inconsistent and not very engaging." authors: ["Pierre-Emmanuel Mazaré", "Samuel Humeau", "Martin Raison", "Antoine Bordes"] year: 2018 venue: "Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing" source: "https://aclanthology.org/D18-1298/" doi: "10.18653/v1/d18-1298" licence: "CC BY 4.0 (ACL Anthology)" licence_url: "https://creativecommons.org/licenses/by/4.0/" summary_licence: "Elamur Cosmic License" canonical: "https://research.elamur.ai/free/en/papers/mazare-2018-training-millions-personalized-dialogue-agents/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/mazare-2018-training-millions-personalized-dialogue-agents/) [home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2018](https://md.research.elamur.ai/free/en/papers/2018/page/2/) / Training Millions of Personalized Dialogue Agents # Training Millions of Personalized Dialogue Agents **Authors:** [Pierre-Emmanuel Mazaré](https://md.research.elamur.ai/free/en/authors/pierre-emmanuel-mazare/), [Samuel Humeau](https://md.research.elamur.ai/free/en/authors/samuel-humeau/), Martin Raison, [Antoine Bordes](https://md.research.elamur.ai/free/en/authors/antoine-bordes/) · **Year:** 2018 · **Venue:** Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/) **Links:** [Publisher's page](https://aclanthology.org/D18-1298/) · [PDF](https://aclanthology.org/D18-1298.pdf) · [DOI 10.18653/v1/d18-1298](https://doi.org/10.18653/v1/d18-1298) · [OpenAlex](https://openalex.org/W2890394457) **Citations:** 226 citations (openalex-snapshot · 2026-10-14) · [108](https://md.research.elamur.ai/free/en/papers/mazare-2018-training-millions-personalized-dialogue-agents/cited-by/) in our library, of the 140884 papers whose bibliographies we read ## Summary End-to-end chit-chat models are often inconsistent and not very engaging. Earlier work showed that giving a model a short textual persona helps. However, the PERSONA-CHAT corpus behind that finding was crowdsourced and contains only about a thousand personas. This paper scales the idea up using a public REDDIT dump of 1.7 billion comments. For each user, the authors collect sentences from that user's own comments that read like self-descriptions, using simple filters on length, first-person words and part-of-speech content. Consecutive comment pairs in a thread become context–response examples. The resulting corpus links roughly 5 million personas to about 700 million dialogues. The models are retrieval-based, ranking candidate answers rather than generating them. Context, response and persona each have their own encoder, and the persona is merged in through a single-hop memory network. Conditioning on personas improves accuracy for bag-of-words, LSTM and Transformer encoders alike. Pretraining on the REDDIT data and then fine-tuning gives a large jump on PERSONA-CHAT. This suggests the corpus is a broadly useful resource for personalized dialogue. ## Key points - Persona sentences must have 4–20 tokens, contain "I" or "my", and include a verb plus a noun, pronoun or adjective. Personas cover 4.6m users with rule-based selection and 7.2m with random selection, out of 13.2m users in total. - On REDDIT (hits@1 among 100 candidates), the Transformer reaches 74.4 with personas versus 69.1 without. The LSTM improves from 63.1 to 66.3, and the bag-of-words encoder from 51.7 to 53.9. - Persona selection matters: - Rule-based selection with up to 100 sentences works best (74.4). - Random sentences from the same user score 73.8. - Random sentences from other users score only 66.9. - Adding a PERSONA-CHAT-trained classifier filter actually hurts. - On PERSONA-CHAT (hits@1 among 20 candidates), the previous best was 35.4. A Transformer trained only on PERSONA-CHAT gets 42.1, and REDDIT pretraining plus fine-tuning reaches 60.7. Without fine-tuning, the REDDIT model scores just 25.6. - Training used Adamax with batch size 512 and FastText initialization. One epoch took about 3 days on 8 GPUs. *Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.* Fields: [Natural language processing](https://md.research.elamur.ai/free/en/shelves/nlp/) Topics: [Dialogue systems](https://md.research.elamur.ai/free/en/topics/dialogue-systems/) ## Cite The paper itself, not this page. BibTeX from the library's record of the work. ```bibtex @inproceedings{mazare2018training, title = {{Training Millions of Personalized Dialogue Agents}}, author = {Mazaré, Pierre-Emmanuel and Humeau, Samuel and Raison, Martin and Bordes, Antoine}, booktitle = {Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing}, year = {2018}, doi = {10.18653/v1/d18-1298}, url = {https://doi.org/10.18653/v1/d18-1298} } ``` - [← previous Predicting Factuality of Reporting and Bias of News Media Sources](https://md.research.elamur.ai/free/en/papers/baly-2018-predicting-factuality-reporting-bias-news/) - [next → TextWorld: A Learning Environment for Text-based Games](https://md.research.elamur.ai/free/en/papers/cote-2018-textworld-learning-environment-text-based/) - [↑ back to papers of 2018](https://md.research.elamur.ai/free/en/papers/2018/page/2/) ## References 5 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet. 1. Antoine Bordes, Y-Lan Boureau, and Jason Weston. 2016. Learning end-to-end goal-oriented dialog. arXiv preprint arXiv:1605.07683. [find](https://scholar.google.com/scholar?q=Antoine+Bordes%2C+Y%2DLan+Boureau%2C+and+Jason+Weston%2E+2016%2E+Learning+end%2Dto%2Dend+goal%2Doriented+dialog%2E+arXiv+preprint+arXiv%3A1605%2E07683%2E) 2. Jesse Dodge, Andreea Gane, Xiang Zhang, Antoine Bordes, Sumit Chopra, Alexander Miller, Arthur Szlam, and Jason Weston. 2015. Evaluating prerequisite qualities for learning end-to-end dialog systems. arXiv preprint arXiv:1511.06931. [find](https://scholar.google.com/scholar?q=Jesse+Dodge%2C+Andreea+Gane%2C+Xiang+Zhang%2C+Antoine+Bordes%2C+Sumit+Chopra%2C+Alexander+Miller%2C+Arthur+Szlam%2C+and+Jason+Weston%2E+2015%2E+Evaluating+prerequisite+qualities+for+learning+end%2Dto%2Dend+dialog+systems%2E+) 3. Marjan Ghazvininejad, Chris Brockett, Ming-Wei Chang, Bill Dolan, Jianfeng Gao, Wen-tau Yih, and Michel Galley. 2017. A knowledge-grounded neural conversation model. CoRR, abs/1702.01932. — [arXiv:1702.01932](https://arxiv.org/abs/1702.01932) [find](https://scholar.google.com/scholar?q=Marjan+Ghazvininejad%2C+Chris+Brockett%2C+Ming%2DWei+Chang%2C+Bill+Dolan%2C+Jianfeng+Gao%2C+Wen%2Dtau+Yih%2C+and+Michel+Galley%2E+2017%2E+A+knowledge%2Dgrounded+neural+conversation+model%2E+CoRR%2C+abs%2F1702%2E01932%2E+%E2%80%94+%5BarXiv%3A170) 4. Chaitanya K Joshi, Fei Mi, and Boi Faltings. 2017. Personalization in goal-oriented dialog. arXiv preprint arXiv:1706.07503. — [arXiv:1706.07503](https://arxiv.org/abs/1706.07503) [find](https://scholar.google.com/scholar?q=Chaitanya+K+Joshi%2C+Fei+Mi%2C+and+Boi+Faltings%2E+2017%2E+Personalization+in+goal%2Doriented+dialog%2E+arXiv+preprint+arXiv%3A1706%2E07503%2E+%E2%80%94+%5BarXiv%3A1706%2E07503%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F1706%2E07503%29) 5. Jiwei Li, Michel Galley, Chris Brockett, Jianfeng Gao, and Bill Dolan. 2016. A persona-based neural conversation model. CoRR, abs/1603.06155. [find](https://scholar.google.com/scholar?q=Jiwei+Li%2C+Michel+Galley%2C+Chris+Brockett%2C+Jianfeng+Gao%2C+and+Bill+Dolan%2E+2016%2E+A+persona%2Dbased+neural+conversation+model%2E+CoRR%2C+abs%2F1603%2E06155%2E) [our card](https://md.research.elamur.ai/free/en/papers/li-2016-persona-based-neural-conversation-model/)