---
title: "Marjan Ghazvininejad — author profile and papers in our library"
description: "Marjan Ghazvininejad, University of Washington: 9 papers on Elamur Research (2017–2023), h-index 26 — affiliations, topics, co-authors and…"
canonical: "https://research.elamur.ai/free/en/authors/marjan-ghazvininejad/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/authors/marjan-ghazvininejad/)

[home](https://md.research.elamur.ai/free/en/) / [authors](https://md.research.elamur.ai/free/en/authors/) / Marjan Ghazvininejad

# Marjan Ghazvininejad

Marjan Ghazvininejad works on neural text generation, with machine translation as the recurring test bed. Their papers here look at how sequence-to-sequence models are pre-trained, how decoding can be made faster, and how generated output can drift away from its source. The methods are mostly Transformer encoder-decoders, denoising and masking objectives, and careful evaluation across languages and domains, touching [language models and pretraining](https://md.research.elamur.ai/free/en/topics/language-models/) and [machine translation](https://md.research.elamur.ai/free/en/topics/machine-translation/).

The most cited work in this group is [BART](https://md.research.elamur.ai/free/en/papers/lewis-2020-bart-denoising-sequence-sequence-pre/) (2020), a sequence-to-sequence model trained to rebuild documents that have been corrupted by arbitrary noise, so that one network serves both understanding and generation. [Mask-Predict](https://md.research.elamur.ai/free/en/papers/ghazvininejad-2019-mask-predict-parallel-decoding-conditional/) (2019) introduced conditional masked language models, which fill in masked target words in parallel and refine them over a few rounds instead of translating word by word. [Pre-training via Paraphrasing](https://md.research.elamur.ai/free/en/papers/lewis-2020-pre-training-paraphrasing/) (2020) replaced masking with retrieval: the MARGE model regenerates a document from related documents, often in other languages, and learns which ones to retrieve from the reconstruction loss itself.

Later papers turn to reliability and to prompting. [Detecting Hallucinated Content in Conditional Neural Sequence Generation](https://md.research.elamur.ai/free/en/papers/zhou-2021-detecting-hallucinated-content-conditional-neural/) (2021) set up a token-level task: mark each output word as supported or unsupported by the source, without needing a reference. [In-context Examples Selection for Machine Translation](https://md.research.elamur.ai/free/en/papers/agrawal-2023-context-examples-selection-machine-translation/) (2023) showed that a single noisy demonstration can badly hurt a prompted translation model, while a well-chosen one helps.

The current affiliation is the [University of Washington](https://md.research.elamur.ai/free/en/universities/university-of-washington/); earlier papers come from [Meta](https://md.research.elamur.ai/free/en/companies/meta/), with groups in the United States, Israel and the United Kingdom, and before that from the [University of Southern California](https://md.research.elamur.ai/free/en/universities/university-of-southern-california/). Frequent co-authors include [Luke Zettlemoyer](https://md.research.elamur.ai/free/en/authors/luke-zettlemoyer/), [Mike Lewis](https://md.research.elamur.ai/free/en/authors/mike-lewis/) and [Chunting Zhou](https://md.research.elamur.ai/free/en/authors/chunting-zhou/).

Marjan Ghazvininejad has 9 papers in our library, in 2017–2023, most of them on [Machine translation](https://md.research.elamur.ai/free/en/topics/machine-translation/), [Language models & pretraining](https://md.research.elamur.ai/free/en/topics/language-models/), [Multilingual & cross-lingual NLP](https://md.research.elamur.ai/free/en/topics/multilingual-nlp/); often with [Luke Zettlemoyer](https://md.research.elamur.ai/free/en/authors/luke-zettlemoyer/), [Mike Lewis](https://md.research.elamur.ai/free/en/authors/mike-lewis/), [Omer Levy](https://md.research.elamur.ai/free/en/authors/omer-levy/).

9 papers in our library · 2017–2023

[OpenAlex](https://openalex.org/A5011974509) · [**h-index 26**](https://openalex.org/authors/A5011974509) · [74 works](https://openalex.org/works?filter=author.id:A5011974509) · [12,493 citations](https://openalex.org/works?filter=author.id:A5011974509&sort=cited_by_count:desc) OpenAlex, 2026-10-14

[📄 9 works in our library](https://md.research.elamur.ai/free/en/authors/marjan-ghazvininejad/#papers) · [🔗 Cited 5,539 times in our library](https://md.research.elamur.ai/free/en/authors/marjan-ghazvininejad/cited-by/) · [2 more in our library without a page yet](#also) · [⭐ Paper Stars ranking: #145](https://md.research.elamur.ai/free/en/paper-stars/page/2/#star-145)

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

Elsewhere: [Google Scholar](https://scholar.google.com/citations?view_op=search_authors&mauthors=Marjan+Ghazvininejad) · [Semantic Scholar](https://www.semanticscholar.org/search?q=Marjan+Ghazvininejad) · [dblp](https://dblp.org/search?q=Marjan+Ghazvininejad) · [ACL Anthology](https://aclanthology.org/search/?q=Marjan+Ghazvininejad)

## Affiliations

[🎓**University of Washington** Seattle · United States 2021–2026 · 5 papers*current*](https://md.research.elamur.ai/free/en/universities/university-of-washington/) · [🏢**Meta** Menlo Park · United States 2019–2023 · 18 papers](https://md.research.elamur.ai/free/en/companies/meta/) · [🎓**University of Southern California** Los Angeles · United States 2015–2019 · 12 papers](https://md.research.elamur.ai/free/en/universities/university-of-southern-california/) · [🏢**Meta** Tel Aviv · Israel 2019–2021 · 6 papers](https://md.research.elamur.ai/free/en/companies/meta-israel/) · [🏢**Meta** London · United Kingdom 2020–2021 · 5 papers](https://md.research.elamur.ai/free/en/companies/meta-united-kingdom/) · [🎓**Carnegie Mellon University** Pittsburgh · United States 2021–2024 · 3 papers](https://md.research.elamur.ai/free/en/universities/carnegie-mellon-university/) · [🎓**Sharif University of Technology** Tehran · Iran 2011 · 2 papers](https://md.research.elamur.ai/free/en/universities/sharif-university-of-technology/)

From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 3 more institutions on single bylines are not listed.

## Fields, topics and concepts

Of its works in our library, with the number of works on each.

📚 Shelves 5 The field of science this work belongs to (OpenAlex fields).[Arts and Humanities · 9 works](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · [Computer Science · 9 works](https://md.research.elamur.ai/free/en/shelves/computer-science/) · [Artificial Intelligence · 9 works](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · [Language and Linguistics · 9 works](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · [Computer Vision and Pattern Recognition · 1 work](https://md.research.elamur.ai/free/en/shelves/computer-science/computer-vision-and-pattern-recognition/)

🧭 Topics 4 Research subjects and languages the work is about.[Text generation · 1 work](https://md.research.elamur.ai/free/en/topics/text-generation/) · [Language models & pretraining · 3 works](https://md.research.elamur.ai/free/en/topics/language-models/) · [Machine translation · 3 works](https://md.research.elamur.ai/free/en/topics/machine-translation/) · [Multilingual & cross-lingual NLP · 2 works](https://md.research.elamur.ai/free/en/topics/multilingual-nlp/)

💡 Concepts 1 Key ideas and methods explained on our concept pages.[Lexicography · 1 work](https://md.research.elamur.ai/free/en/concepts/lexicography/)

## Works by year

### 2020s

[**2023** 3 papers Marjan Ghazvininejad's papers of 2023](https://md.research.elamur.ai/free/en/authors/marjan-ghazvininejad/year/2023/) · [**2021** 1 paper Marjan Ghazvininejad's papers of 2021](https://md.research.elamur.ai/free/en/papers/zhou-2021-detecting-hallucinated-content-conditional-neural/) · [**2020** 2 papers Marjan Ghazvininejad's papers of 2020](https://md.research.elamur.ai/free/en/authors/marjan-ghazvininejad/year/2020/)

### 2010s

[**2019** 2 papers Marjan Ghazvininejad's papers of 2019](https://md.research.elamur.ai/free/en/authors/marjan-ghazvininejad/year/2019/) · [**2017** 1 paper Marjan Ghazvininejad's papers of 2017](https://md.research.elamur.ai/free/en/papers/ghazvininejad-2017-hafez-interactive-poetry-generation-system/)

## Papers

The most cited first (OpenAlex and our library).

1. **[BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://md.research.elamur.ai/free/en/papers/lewis-2020-bart-denoising-sequence-sequence-pre/)** *(2020 · 12,784 citations)* The paper addresses a gap in pretraining. BERT-style masked models read context in both directions but are hard to use for generation, while GPT-style left-to-right models can generate but see only preceding context. The authors propose BART, a standard Transformer encoder–decoder pretrained as a denoising…
2. **[Mask-Predict: Parallel Decoding of Conditional Masked Language Models](https://md.research.elamur.ai/free/en/papers/ghazvininejad-2019-mask-predict-parallel-decoding-conditional/)** *(2019 · 671 citations)* Translation models usually produce output one token at a time, left to right, so decoding time grows with sentence length.
3. **[In-context Examples Selection for Machine Translation](https://md.research.elamur.ai/free/en/papers/agrawal-2023-context-examples-selection-machine-translation/)** *(2023 · 242 citations)* The paper asks how the choice and ordering of in-context demonstrations shape the output of a large generative model when it is used for machine translation.
4. **[Detecting Hallucinated Content in Conditional Neural Sequence Generation](https://md.research.elamur.ai/free/en/papers/zhou-2021-detecting-hallucinated-content-conditional-neural/)** *(2021 · 229 citations)* Neural generators for translation and summarization can produce fluent output that contains material the input does not support.
5. **[Hafez: an Interactive Poetry Generation System](https://md.research.elamur.ai/free/en/papers/ghazvininejad-2017-hafez-interactive-poetry-generation-system/)** *(2017 · 196 citations)* Hafez is a demo system that writes English sonnet-style poems on any topic a user supplies. It targets three weaknesses of earlier poetry generators: poem quality is hard to evaluate, users cannot steer a draft beyond changing the random seed, and generation is slow. The pipeline builds on the authors' earlier work.…
6. **[Pre-training via Paraphrasing](https://md.research.elamur.ai/free/en/papers/lewis-2020-pre-training-paraphrasing/)** *(2020 · 181 citations)* Almost all large-scale NLP pre-training relies on masked language modeling, where parts of a text are hidden and then restored.
7. **[Training on Synthetic Noise Improves Robustness to Natural Noise in Machine Translation](https://md.research.elamur.ai/free/en/papers/karpukhin-2019-training-synthetic-noise-improves-robustness/)** *(2019 · 159 citations)* Translation systems trained on clean text tend to fail on inputs with typos. This holds even for subword (BPE) and character-CNN encoders, because a single misspelled character changes the units these models depend on. Earlier attempts to fix this with synthetic noise only helped on the same artificial noise and hurt…
8. **[XLM-V: Overcoming the Vocabulary Bottleneck in Multilingual Masked Language Models](https://md.research.elamur.ai/free/en/papers/liang-2023-xlm-v-overcoming-vocabulary-bottleneck/)** *(2023 · 117 citations)* Multilingual encoders such as XLM-R have grown in depth and parameter count, yet their vocabularies have stayed near 250K tokens shared by more than 100 languages.
9. **[Dictionary-based Phrase-level Prompting of Large Language Models for Machine Translation](https://md.research.elamur.ai/free/en/papers/ghazvininejad-2023-dictionary-based-phrase-level-prompting/)** *(2023 · 95 citations)* Prompted LLMs translate surprisingly well without MT-specific training, but they often mistranslate rare words, which are common in low-resource language pairs and out-of-domain text.

## All works

11 works of theirs in our library, the most cited first; 9 with a page here, the others link their original until we summarise them.

1. **[BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://md.research.elamur.ai/free/en/papers/lewis-2020-bart-denoising-sequence-sequence-pre/) [Annual Meeting of the Association for Computational Linguistics (ACL)](https://md.research.elamur.ai/free/en/venues/acl/)**Mike Lewis, Yinhan Liu, Naman Goyal, et al.*2020 · 12,784 citations*
2. **[Mask-Predict: Parallel Decoding of Conditional Masked Language Models](https://md.research.elamur.ai/free/en/papers/ghazvininejad-2019-mask-predict-parallel-decoding-conditional/) [Conference on Empirical Methods in Natural Language Processing (EMNLP)](https://md.research.elamur.ai/free/en/venues/emnlp/)**Marjan Ghazvininejad, Omer Levy, Yinhan Liu, et al.*2019 · 671 citations*
3. **[In-context Examples Selection for Machine Translation](https://md.research.elamur.ai/free/en/papers/agrawal-2023-context-examples-selection-machine-translation/) [Findings of the Association for Computational Linguistics](https://md.research.elamur.ai/free/en/venues/findings/)**Sweta Agrawal, Chunting Zhou, Mike Lewis, et al.*2023 · 242 citations*
4. **[Detecting Hallucinated Content in Conditional Neural Sequence Generation](https://md.research.elamur.ai/free/en/papers/zhou-2021-detecting-hallucinated-content-conditional-neural/) [Findings of the Association for Computational Linguistics](https://md.research.elamur.ai/free/en/venues/findings/)**Chunting Zhou, Graham Neubig, Jiatao Gu, et al.*2021 · 229 citations*
5. **[Hafez: an Interactive Poetry Generation System](https://md.research.elamur.ai/free/en/papers/ghazvininejad-2017-hafez-interactive-poetry-generation-system/) [Annual Meeting of the Association for Computational Linguistics (ACL)](https://md.research.elamur.ai/free/en/venues/acl/)**Marjan Ghazvininejad, Xing Shi, Jay Priyadarshi, et al.*2017 · 196 citations*
6. [**Pre-training via Paraphrasing arXiv:2006.15020**Mike Lewis, Marjan Ghazvininejad, Gargi Ghosh, et al.*2020 · 181 citations*](https://md.research.elamur.ai/free/en/papers/lewis-2020-pre-training-paraphrasing/)
7. **[Training on Synthetic Noise Improves Robustness to Natural Noise in Machine Translation](https://md.research.elamur.ai/free/en/papers/karpukhin-2019-training-synthetic-noise-improves-robustness/) [Workshop on Noisy User-generated Text](https://md.research.elamur.ai/free/en/venues/wnut/)**Vladimir Karpukhin, Omer Levy, Jacob Eisenstein, et al.*2019 · 159 citations*
8. **[A Review on Language Models as Knowledge Bases ↗](https://arxiv.org/abs/2204.06031) arXiv:2204.06031 · no summary yet**Badr AlKhamissi et al.*2022 · 148 citations*
9. **[XLM-V: Overcoming the Vocabulary Bottleneck in Multilingual Masked Language Models](https://md.research.elamur.ai/free/en/papers/liang-2023-xlm-v-overcoming-vocabulary-bottleneck/) [Conference on Empirical Methods in Natural Language Processing (EMNLP)](https://md.research.elamur.ai/free/en/venues/emnlp/)**Davis Liang, Hila Gonen, Yuning Mao, et al.*2023 · 117 citations*
10. **[Natural Language to Code Translation with Execution ↗](https://aclanthology.org/2022.emnlp-main.231/) Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing · no summary yet**Freda Shi et al.*2022 · 113 citations*
11. [**Dictionary-based Phrase-level Prompting of Large Language Models for Machine Translation arXiv:2302.07856**Marjan Ghazvininejad, Hila Gonen, Luke Zettlemoyer*2023 · 95 citations*](https://md.research.elamur.ai/free/en/papers/ghazvininejad-2023-dictionary-based-phrase-level-prompting/)

## Co-authors here

12 of 24 co-authors with a page, the most papers together first.

- [Luke Zettlemoyer](https://md.research.elamur.ai/free/en/authors/luke-zettlemoyer/) ×7
- [Mike Lewis](https://md.research.elamur.ai/free/en/authors/mike-lewis/) ×3
- [Omer Levy](https://md.research.elamur.ai/free/en/authors/omer-levy/) ×3
- [Chunting Zhou](https://md.research.elamur.ai/free/en/authors/chunting-zhou/) ×2
- [Hila Gonen](https://md.research.elamur.ai/free/en/authors/hila-gonen/) ×2
- [Naman Goyal](https://md.research.elamur.ai/free/en/authors/naman-goyal/) ×2
- [Yinhan Liu](https://md.research.elamur.ai/free/en/authors/yinhan-liu/) ×2
- [Abdelrahman Mohamed](https://md.research.elamur.ai/free/en/authors/abdelrahman-mohamed/)
- [Armen Aghajanyan](https://md.research.elamur.ai/free/en/authors/armen-aghajanyan/)
- [Davis Liang](https://md.research.elamur.ai/free/en/authors/davis-liang/)
- [Francisco Guzmán](https://md.research.elamur.ai/free/en/authors/francisco-guzman/)
- [Graham Neubig](https://md.research.elamur.ai/free/en/authors/graham-neubig/)

## Venues

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

- [Conference on Empirical Methods in Natural Language Processing (EMNLP)](https://md.research.elamur.ai/free/en/venues/emnlp/) ×3: [EMNLP 2023](https://md.research.elamur.ai/free/en/venues/emnlp/2023/), [EMNLP 2019](https://md.research.elamur.ai/free/en/venues/emnlp/2019/)
- [Annual Meeting of the Association for Computational Linguistics (ACL)](https://md.research.elamur.ai/free/en/venues/acl/) ×2: [ACL 2020](https://md.research.elamur.ai/free/en/venues/acl/2020/), [ACL 2017](https://md.research.elamur.ai/free/en/venues/acl/2017/)
- [Findings of the Association for Computational Linguistics](https://md.research.elamur.ai/free/en/venues/findings/) ×2: [Findings of the Association for Computational Linguistics 2023](https://md.research.elamur.ai/free/en/venues/findings/2023/), [Findings of the Association for Computational Linguistics 2021](https://md.research.elamur.ai/free/en/venues/findings/2021/)
- [Workshop on Noisy User-generated Text](https://md.research.elamur.ai/free/en/venues/wnut/): [Workshop on Noisy User-generated Text 2019](https://md.research.elamur.ai/free/en/venues/wnut/2019/)

## Cited by

Papers of our library cite their works 5,539 times (of the 141,840 papers whose bibliographies we read); [418 citing papers with a page here →](https://md.research.elamur.ai/free/en/authors/marjan-ghazvininejad/cited-by/).

Updated 24 October 2026