Yale University
Nal KalchbrennerFounding Member · PrometheusFounder↗
Researcher
Rewon Child
PrometheusFounding Member of Technical Staff
Child co-created Sparse Transformers, which cut the memory and computation needed for models to connect distant parts of text, images and audio.[1] He also contributed to GPT-2, GPT-3 and Image GPT, helping show how learning from large amounts of unlabelled data could produce versatile language and image models.[2]
Career & education
Current rolePrometheusJan 2026-presentFounding Member of Technical Staff
PrometheusJan 2026-presentPreviously
Microsoft AI2024-2025Member of Technical Staff
Inflection AI2022-2024Founding team
Google Brain2021-2022Researcher
OpenAI2018-2022Member of Technical Staff
Baidu2016-2017ML Research
Microsoft AI2024-2025Inflection AI2022-2024Google Brain2021-2022OpenAI2018-2022Baidu2016-2017Education
Work and education history is primarily focused on AI-relevant roles and may not be comprehensive.
Citation Trend
Total citations to scholarly work, including research papers, technical reports, books and theses, as recorded by Google Scholar. Each point is a dated snapshot.
citations173,911As of Nov 1, '26
citations+4,576Since last snapshot, Oct 1, '26
Selected contributions
Language Models are Few-Shot Learners
With Tom Brown, Ben Mann, Nick Ryder, Melanie Subbiah, Jared Kaplan, Prafulla Dhariwal, Arvind Neelakantan and others ·GPT-3 performs tasks from prompts and a few examples without task-specific training.
Language Models are Unsupervised Multitask Learners
GPT-2 showed how learning to predict text could support tasks without separate training.
PaLM: Scaling Language Modeling with Pathways
With Aakanksha Chowdhery, Sharan Narang, Jacob Devlin, Maarten Bosma, Gaurav Mishra, Adam Roberts, Paul Barham and others ·Presented PaLM, a large language model demonstrating gains in reasoning, translation and coding.
References
- Generating Long Sequences with Sparse Transformersarxiv.org
Child co-created Sparse Transformers, which cut the memory and computation needed for models to connect distant parts of text, images and audio.
- Rewon Child — research contributionsrewonc.github.io
He also contributed to GPT-2, GPT-3 and Image GPT, helping show how learning from large amounts of unlabelled data could produce versatile language and image models.
Based on public sources. If accuracy is important, verify details using the links or references.
For corrections, updates, or requests to remove information, contact us.
Reviewed August 7, 2026 · Checked for accuracy and new information.
