🧑‍🔬 Quoc Viet Le

🏛️ Affiliations

🏢GoogleMountain View · United States2012–2024 · 144 papers🎓Stanford UniversityStanford · United States2007–2024 · 26 papers🎓Australian National UniversityCanberra · Australia2005–2008 · 9 papers🔬Data61Sydney · Australia2005–2008 · 8 papers🎓University of Massachusetts AmherstAmherst Center · United States2015–2024 · 4 papers🎓Tel Aviv UniversityTel Aviv · Israel2016–2017 · 3 papers🎓Carnegie Mellon UniversityPittsburgh · United States2019–2026 · 2 papers🎓University of California, BerkeleyBerkeley · United States2022–2026 · 2 papers

From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 7 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.

📅 Works by year

2020s

2010s

📄 Papers

The most cited first (OpenAlex and our library).

1 2 next →

1. Sequence to Sequence Learning with Neural Networks (2014 · 18,409 citations)
   Standard deep networks expect inputs and outputs of fixed size, but tasks like translation pair sequences whose lengths are unknown and differ.
2. Chain-of-Thought Prompting Elicits Reasoning in Large Language Models (2022 · 11,499 citations)
   Large language models handle many tasks well through few-shot prompting, but making them bigger has not been enough for multi-step problems such as arithmetic, commonsense and symbolic reasoning.
3. Learning Transferable Architectures for Scalable Image Recognition (2017 · 6,583 citations)
   Hand-designing convolutional networks for image classification takes a great deal of expert effort. Searching for architectures automatically with Neural Architecture Search (NAS) is very expensive on a dataset as large as ImageNet. Zoph et al. avoid this cost by searching on the small CIFAR-10 dataset and then…
4. Distributed Representations of Sentences and Documents (2014 · 5,803 citations)
   Classifiers and clustering methods need each text as a fixed-length vector. The usual choice, bag-of-words, discards word order and treats every pair of words as equally unrelated. Le and Mikolov propose Paragraph Vector, an unsupervised way to learn a dense vector for a text of any length, from a phrase to a full…
5. EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks (2019 · 5,416 citations)
   The paper asks whether there is a principled way to enlarge a convolutional network when more compute becomes available.
6. Neural Architecture Search with Reinforcement Learning (2016 · 4,413 citations)
   Building a good neural network by hand still takes a lot of expert knowledge and trial and error. Zoph and Le treat architecture design as a learning problem. In their Neural Architecture Search, a recurrent "controller" writes out a network description as a sequence of tokens, such as filter sizes, strides and…
7. XLNet: Generalized Autoregressive Pretraining for Language Understanding (2019 · 3,857 citations)
   Pretraining for language understanding had split into two camps. Autoregressive language models such as GPT estimate text likelihood with the product rule, but they see context from only one side. BERT-style denoising autoencoders see both sides. However, they train on inputs containing [MASK] symbols that never…
8. SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition (2019 · 3,837 citations)
   End-to-end speech recognizers overfit easily and need large amounts of training data. Earlier augmentation for ASR mostly changed the raw audio, for example by adding noise, perturbing speed or simulating rooms. The authors propose SpecAugment, which instead alters the log mel spectrogram, treating it like an image…
9. Self-Consistency Improves Chain of Thought Reasoning in Language Models (2022 · 3,121 citations)
   Chain-of-thought prompting lets large language models write out intermediate reasoning before answering. The usual setup, however, decodes only one greedy path, so a single faulty step spoils the result. The authors replace greedy decoding with a "sample-and-marginalize" procedure they call self-consistency. They…
10. Google's Multilingual Neural Machine Translation System: Enabling Zero-Shot Translation (2017 · 2,630 citations)
    The authors from Google address the task of translating between many languages with a single neural model (NMT).
11. Finetuned Language Models Are Zero-Shot Learners (2021 · 1,954 citations)
    Large language models such as GPT-3 perform well when the prompt contains a few examples. Without examples, that is, in the zero-shot setting, their quality drops noticeably. The authors from Google Research propose instruction tuning: a pretrained model is fine-tuned on a mixture of tasks, each formulated as an…
12. Unsupervised Data Augmentation for Consistency Training (2019 · 1,871 citations)
    Deep models usually need many labeled examples, and semi-supervised methods based on consistency training try to reduce that need.
13. A Neural Conversational Model (2015 · 1,791 citations)
    The paper asks whether a dialogue agent can be built without hand-written rules or a narrow domain by treating conversation as plain sequence prediction.
14. Scaling Up Visual and Vision-Language Representation Learning With Noisy Text Supervision (2021 · 1,615 citations)
    The paper asks whether visual and vision-language representations can be learned the way NLP models learn from raw web text, without expensive curation or labels.
15. Program Synthesis with Large Language Models (2021 · 1,192 citations)
    The paper asks how far general-purpose large Transformer language models can go at writing short programs in an ordinary language like Python, rather than in a narrow DSL.
16. Challenging BIG-Bench Tasks and Whether Chain-of-Thought Can Solve Them (2023 · 1,119 citations)
    BIG-Bench was designed to probe abilities supposedly beyond current language models, yet PaLM 540B already beat the average human rater on most of its tasks.
17. Grounded Compositional Semantics for Finding and Describing Images with Sentences (2014 · 932 citations)
    The paper addresses how to build sentence representations that reflect the visual scene a sentence describes, so that sentences and images can be retrieved from each other.
18. Addressing the Rare Word Problem in Neural Machine Translation (2015 · 832 citations)
    Neural machine translation (NMT) systems of this period kept their output vocabulary small, often the 30K–80K most frequent words, because the softmax is expensive. Every out-of-vocabulary word therefore became a single unk symbol, and sentences full of names, numbers and rare terms came out badly.
19. The Flan Collection: Designing Data and Methods for Effective Instruction Tuning (2023 · 607 citations)
    Several public instruction-tuning collections already existed, each using different tasks, model sizes and prompt formats.
20. QANet: Combining Local Convolution with Global Self-Attention for Reading Comprehension (2018 · 597 citations)
    In 2018, leading reading-comprehension systems for SQuAD were built on recurrent networks combined with attention.
21. Towards a Human-like Open-Domain Chatbot (2020 · 592 citations)
    Open-domain chatbots have tended to give replies that are either nonsensical or bland, and the stronger systems relied on hand-built pipelines with dialog managers and rule or retrieval components.
22. Massive Exploration of Neural Machine Translation Architectures (2017 · 574 citations)
    NMT models are costly to train, so researchers tend to reuse familiar architectures without knowing which design decisions actually drive translation quality.
23. Solving olympiad geometry without human demonstrations (2024 · 498 citations)
    Automated proving of olympiad-level geometry has been limited by a lack of training data. Few human proofs exist in a machine-readable form, and translating them into formal language is expensive. The authors present AlphaGeometry, a theorem prover trained without any human demonstrations. They sample random…
24. Semi-Supervised Sequence Modeling with Cross-View Training (2018 · 494 citations)
    Pre-trained representations such as word2vec or ELMo are learned from unlabeled text without any knowledge of the target task. The labeled data only enters later, during supervised training. Clark et al. propose Cross-View Training (CVT), a semi-supervised method that mixes both kinds of data throughout training.
25. Large Language Models as Optimizers (2023 · 413 citations)
    Many practical optimization problems have no usable gradient, and derivative-free methods usually need custom engineering for each task.

1 2 next →

All works

243 works of theirs in our library, the most cited first; 34 with a page here, the others link their original until we summarise them.

1. Sequence to Sequence Learning with Neural NetworksarXiv:1409.3215Ilya Sutskever, Oriol Vinyals, Quoc V. Le2014 · 18,409 citations
2. Chain-of-Thought Prompting Elicits Reasoning in Large Language ModelsarXiv:2201.11903Jason Wei, Xuezhi Wang, Dale Schuurmans, et al.2022 · 11,499 citations
3. Searching for MobileNetV3 ↗IEEE/CVF International Conference on Computer Vision (ICCV) · no summary yetAndrew Howard et al.2019 · 9,633 citations
4. EfficientDet: Scalable and Efficient Object Detection ↗IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · no summary yetMingxing Tan, Ruoming Pang, Quoc Viet Le2020 · 9,341 citations
5. Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation ↗arXiv:1609.08144 · no summary yetYonghui Wu et al.2016 · 6,904 citations
6. Learning Transferable Architectures for Scalable Image RecognitionIEEE Computer Society Conference on Computer Vision and Pattern RecognitionBarret Zoph, Vijay Vasudevan, Jonathon Shlens, et al.2017 · 6,583 citations
7. Distributed Representations of Sentences and DocumentsarXiv:1405.4053Quoc V. Le, Tomas Mikolov2014 · 5,803 citations
8. EfficientNet: Rethinking Model Scaling for Convolutional Neural NetworksInternational Conference on Machine Learning (ICML)Mingxing Tan, Quoc V. Le2019 · 5,416 citations
9. Neural Architecture Search with Reinforcement LearningarXiv:1611.01578Barret Zoph, Quoc V. Le2016 · 4,413 citations
10. Transformer-XL: Attentive Language Models beyond a Fixed-Length Context ↗Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics · no summary yetZihang Dai et al.2019 · 3,901 citations
11. XLNet: Generalized Autoregressive Pretraining for Language UnderstandingarXiv:1906.08237Zhilin Yang, Zihang Dai, Yiming Yang, et al.2019 · 3,857 citations
12. SpecAugment: A Simple Data Augmentation Method for Automatic Speech RecognitionInterspeechDaniel S. Park, William Chan, Yu Zhang, et al.2019 · 3,837 citations
13. MnasNet: Platform-Aware Neural Architecture Search for Mobile ↗IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · no summary yetMingxing Tan et al.2019 · 3,454 citations
14. Scaling Instruction-Finetuned Language Models ↗arXiv:2210.11416 · no summary yetHyung Won Chung et al.2022 · 3,392 citations
15. Randaugment: Practical automated data augmentation with a reduced search space ↗IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · no summary yetEkin D. Cubuk et al.2020 · 3,233 citations
16. Self-Consistency Improves Chain of Thought Reasoning in Language ModelsarXiv:2203.11171Xuezhi Wang, Jason Wei, Dale Schuurmans, et al.2022 · 3,121 citations
17. Large Scale Distributed Deep Networks ↗no summary yetJay B. Dean et al.2012 · 3,047 citations
18. AutoAugment: Learning Augmentation Strategies From Data ↗IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · no summary yetEkin D. Cubuk et al.2019 · 2,810 citations
19. Google's Multilingual Neural Machine Translation System: Enabling Zero-Shot TranslationTransactions of the Association for Computational Linguistics (TACL)Melvin Johnson, Mike Schuster, Quoc V. Le, et al.2017 · 2,630 citations
20. Scalable and accurate deep learning with electronic health records ↗npj Digital Medicine · no summary yetAlvin Rajkomar et al.2018 · 2,607 citations
21. Self-Training With Noisy Student Improves ImageNet Classification ↗IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · no summary yetQizhe Xie et al.2020 · 2,536 citations
22. Listen, attend and spell: A neural network for large vocabulary conversational speech recognition ↗IEEE International Conference on Acoustics Speech and Signal Processing · no summary yetWilliam Chan et al.2016 · 2,374 citations
23. NAS-FPN: Learning Scalable Feature Pyramid Architecture for Object Detection ↗IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · no summary yetGolnaz Ghiasi, Tsung-Yi Lin, Quoc Viet Le2019 · 2,047 citations
24. Finetuned Language Models Are Zero-Shot LearnersarXiv:2109.01652Jason Wei, Maarten Bosma, Vincent Y. Zhao, et al.2021 · 1,954 citations
25. Unsupervised Data Augmentation for Consistency TrainingarXiv:1904.12848Qizhe Xie, Zihang Dai, Eduard Hovy, et al.2019 · 1,871 citations

See all 243 works by Quoc Viet Le →

🤝 Co-authors here

12 of 109 co-authors with a page, the most papers together first.

• Minh-Thang Luong ×7
• Denny Zhou ×6
• Jason Wei ×6
• Barret Zoph ×4
• Oriol Vinyals ×4
• Xuezhi Wang ×4
• Christopher D. Manning ×3
• Dale Schuurmans ×3
• Ed Chi ×3
• Ilya Sutskever ×3
• Maarten Bosma ×3
• Thang Luong ×3

Venues

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

• IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings ×14
• International Conference on Machine Learning (ICML) ×11: ICML 2021, ICML 2019
• Conference on Empirical Methods in Natural Language Processing (EMNLP) ×9: EMNLP 2018, EMNLP 2017 ×2
• Annual Meeting of the Association for Computational Linguistics (ACL) ×6: ACL 2019, ACL 2015
• Lecture notes in computer science ×6
• neural information processing systems ×6
• International Conference on Learning Representations ×5
• Interspeech ×4: Interspeech 2019
• IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) ×4
• Nature ×3: Nature 2024
• Transactions of the Association for Computational Linguistics (TACL) ×3: Transactions of the Association for Computational Linguistics (TACL) 2017, Transactions of the Association for Computational Linguistics (TACL) 2014
• IEEE International Conference on Acoustics Speech and Signal Processing ×3

🔗 Cited by

Papers of our library cite their works 47,619 times (of the 149,439 papers whose bibliographies we read); 4,559 citing papers with a page here →.

Updated 30 October 2026

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/authors/quoc-viet-le/