Elamur Research⌕
☰
home / authors / Margaret Li

Margaret Li

Margaret Li has 2 papers in our library, in 2020–2021, most of them on Dialogue systems, Bias, fairness & safety; often with Jason Weston, Y-Lan Boureau, Da Ju.

2 papers in our library · 2020–2021

OpenAlex · ORCID · Wikidata · h-index 13 · 33 works · 1,589 citations OpenAlex, 2026-10-14

📄 2 works in our library · 🔗 Cited 443 times in our library · 4 more in our library without a page yet · ⭐ Paper Stars ranking: #3943

The profile is the library's match of the name; two people with one name are not always told apart.

Elsewhere: Google Scholar · Semantic Scholar · dblp · ACL Anthology

 Affiliations

🎓University of WashingtonSeattle · United States2022–2025 · 11 papers🏢MetaMenlo Park · United States2019–2023 · 7 papers🏢MetaLondon · United Kingdom2019–2021 · 4 papers🤝Institute for Systems BiologySeattle · United States2023 · 3 papers

From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 6 more institutions on single bylines are not listed.

 Concepts, shelves and topics

Of its works in our library: the number of works on each and its rating among them.

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Artificial Intelligence · 2 worksArts and Humanities · 2 worksComputer Science · 2 worksLanguage and Linguistics · 2 works

🧭 Topics 2Research subjects and languages the work is about.Dialogue systems · 2 worksBias, fairness & safety · 1 work

 Works by year

2020s

20211 paperBot-Adversarial Dialogue for Safe Conversational Agents20201 paperDon't Say That! Making Inconsistent Dialogue Unlikely with Unlikelihood Training

 Papers

The most cited first (OpenAlex and our library).

  1. Don't Say That! Making Inconsistent Dialogue Unlikely with Unlikelihood Training (2020 · 215 citations)
    Generative dialogue models trained only with maximum likelihood show several weaknesses. They copy too much from the context, repeat n-grams inside a single reply, overuse frequent tokens, and can contradict what was said earlier. The authors extend the unlikelihood loss of Welleck et al. (2019a) into a general…
  2. Bot-Adversarial Dialogue for Safe Conversational Agents (2021 · 139 citations)
    Open-domain chatbots trained on large human conversation corpora can pick up toxic behavior. Most safety work has studied human-written text, not what bots say when users deliberately provoke them. The authors propose Bot-Adversarial Dialogue (BAD), a human-and-model-in-the-loop framework. Crowdworkers chat with…

All works

6 works of theirs in our library, the most cited first; 2 with a page here, the others link their original until we summarise them.

  1. Don't Say That! Making Inconsistent Dialogue Unlikely with Unlikelihood TrainingAnnual Meeting of the Association for Computational Linguistics (ACL)Margaret Li, Stephen Roller, Ilia Kulikov, et al.2020 · 215 citations
  2. Recipes for Safety in Open-domain ChatbotsarXiv:2010.07079 · original ↗ · no summary yetJing Xu et al.2020 · 198 citations
  3. Bot-Adversarial Dialogue for Safe Conversational AgentsConference of the North American Chapter of the Association for Computational Linguistics (NAACL)Jing Xu, Da Ju, Margaret Li, et al.2021 · 139 citations
  4. Branch-Train-Merge: Embarrassingly Parallel Training of Expert Language ModelsarXiv:2208.03306 · original ↗ · no summary yetMargaret Li et al.2022 · 108 citations
  5. In-context Pretraining: Language Modeling Beyond Document BoundariesarXiv:2310.10638 · original ↗ · no summary yetWeijia Shi et al.2023 · 55 citations
  6. Breaking the Curse of Multilinguality with Cross-lingual Expert Language ModelsProceedings of the 2024 Conference on Empirical Methods in Natural Language Processing · original ↗ · no summary yetTerra Blevins et al.2024 · 41 citations

 Co-authors here

9 co-authors with a page, the most papers together first.

Venues

Where their works here were published, the most works first, with the editions of those that have a page.

 Cited by

Papers of our library cite their works 443 times (of the 141,840 papers whose bibliographies we read); 63 citing papers with a page here →.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/authors/margaret-li/