--- title: "David R. So — author profile and papers" description: "David R. So, Microsoft: 3 papers on Elamur Research (2020–2022), h-index 12 — affiliations, topics, co-authors and who cites them in our library." canonical: "https://research.elamur.ai/free/en/authors/david-r-so/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/authors/david-r-so/) [home](https://md.research.elamur.ai/free/en/) / [authors](https://md.research.elamur.ai/free/en/authors/) / David R. So # David R. So David R. So works on large neural network models for natural language. The papers here, from 2020 and 2021, take up two related questions: whether a single generative model trained end to end can hold an open-ended conversation, and how much energy and carbon the training of such models really costs. The methods range from training Transformer-based models on very large conversational corpora and scoring their replies with crowd-sourced human judgements, to estimating power use from hardware details, datacenter efficiency and the local electricity mix. The first strand belongs to the study of [Dialogue systems](https://md.research.elamur.ai/free/en/topics/dialogue-systems/). The most cited of these papers, [Towards a Human-like Open-Domain Chatbot](https://md.research.elamur.ai/free/en/papers/adiwardana-2020-towards-human-like-open-domain/) (2020), presents Meena. Meena is a 2.6B-parameter seq2seq model built on the Evolved Transformer, and its sole training goal is predicting the next token of filtered social media conversations. The paper also proposes Sensibleness and Specificity Average (SSA), a human evaluation in which raters first check whether a reply fits its context and then whether it is specific to it. The later [Carbon Emissions and Large Neural Network Training](https://md.research.elamur.ai/free/en/papers/patterson-2021-carbon-emissions-large-neural-network/) (2021) turns to the cost of this kind of work. It estimates training energy and CO2e for GPT-3, T5, Switch Transformer, GShard and Meena. It also re-examines an earlier figure for the architecture search behind the Evolved Transformer and finds that figure much too high. Taken together, the two papers move from building a large dialogue model to accounting for the resources such training consumes. So's bylines list [Google](https://md.research.elamur.ai/free/en/companies/google/) in Mountain View from 2017 to 2022, and [Microsoft](https://md.research.elamur.ai/free/en/companies/microsoft/) in Redmond as the current institution, from 2025. Co-authors on the papers here include [Quoc V. Le](https://md.research.elamur.ai/free/en/authors/quoc-v-le/), on two of them, along with [Jeff Dean](https://md.research.elamur.ai/free/en/authors/jeff-dean/), [Minh-Thang Luong](https://md.research.elamur.ai/free/en/authors/minh-thang-luong/), [Romal Thoppilan](https://md.research.elamur.ai/free/en/authors/romal-thoppilan/) and [Noah Fiedel](https://md.research.elamur.ai/free/en/authors/noah-fiedel/). David R. So has 3 papers in our library, in 2020–2022, most of them on [Dialogue systems](https://md.research.elamur.ai/free/en/topics/dialogue-systems/); often with [Quoc V. Le](https://md.research.elamur.ai/free/en/authors/quoc-v-le/), [Aakanksha Chowdhery](https://md.research.elamur.ai/free/en/authors/aakanksha-chowdhery/), [Apoorv Kulshreshtha](https://md.research.elamur.ai/free/en/authors/apoorv-kulshreshtha/). 3 papers in our library · 2020–2022 [OpenAlex](https://openalex.org/A5067468560) · [**h-index 12**](https://openalex.org/authors/A5067468560) · [22 works](https://openalex.org/works?filter=author.id:A5067468560) · [1,440 citations](https://openalex.org/works?filter=author.id:A5067468560&sort=cited_by_count:desc) OpenAlex, 2026-10-14 [📄 3 works in our library](https://md.research.elamur.ai/free/en/authors/david-r-so/#papers) · [🔗 1,371 citations in our library](https://md.research.elamur.ai/free/en/authors/david-r-so/cited-by/) · [⭐ Paper Star rank #2108](https://md.research.elamur.ai/free/en/paper-stars/page/22/#star-2108) 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=David+R.+So) · [Semantic Scholar](https://www.semanticscholar.org/search?q=David+R.+So) · [dblp](https://dblp.org/search?q=David+R.+So) · [ACL Anthology](https://aclanthology.org/search/?q=David+R.+So) ## Affiliations [🏢**Microsoft** Redmond · United States 2025 · 1 paper*current*](https://md.research.elamur.ai/free/en/companies/microsoft/) · [🏢**Google** Mountain View · United States 2017–2022 · 11 papers](https://md.research.elamur.ai/free/en/companies/google/) From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 1 more institution on single bylines are not listed. ## Topics [Dialogue systems (1)](https://md.research.elamur.ai/free/en/topics/dialogue-systems/) ## Works by year ### 2020s [**2022** 1 paper](https://md.research.elamur.ai/free/en/papers/tay-2022-transcending-scaling-laws-0-1/) · [**2021** 1 paper](https://md.research.elamur.ai/free/en/papers/patterson-2021-carbon-emissions-large-neural-network/) · [**2020** 1 paper](https://md.research.elamur.ai/free/en/papers/adiwardana-2020-towards-human-like-open-domain/) ## Papers The most cited first (OpenAlex and our library). 1. **[Towards a Human-like Open-Domain Chatbot](https://md.research.elamur.ai/free/en/papers/adiwardana-2020-towards-human-like-open-domain/)** *(2020 · 573 citations)* Open-domain chatbots at the time were either built on elaborate hand-engineered pipelines or tended to give replies that made no sense or said nothing in particular. 2. **[Carbon Emissions and Large Neural Network Training](https://md.research.elamur.ai/free/en/papers/patterson-2021-carbon-emissions-large-neural-network/)** *(2021 · 304 citations)* The paper asks how much electricity and greenhouse-gas output (CO2e) the training of large NLP models actually requires. 3. **[Transcending Scaling Laws with 0.1% Extra Compute](https://md.research.elamur.ai/free/en/papers/tay-2022-transcending-scaling-laws-0-1/)** *(2022 · 40 citations)* Making large language models better usually means scaling them up, which costs a great deal of compute. The authors ask whether an already trained causal model can be improved cheaply instead. Their method, UL2R (UL2Restore), takes existing PaLM checkpoints and trains them for a small number of extra steps using the… ## Co-authors here 12 of 20 co-authors with a page, the most papers together first. - [Quoc V. Le](https://md.research.elamur.ai/free/en/authors/quoc-v-le/) ×2 - [Aakanksha Chowdhery](https://md.research.elamur.ai/free/en/authors/aakanksha-chowdhery/) - [Apoorv Kulshreshtha](https://md.research.elamur.ai/free/en/authors/apoorv-kulshreshtha/) - [Denny Zhou](https://md.research.elamur.ai/free/en/authors/denny-zhou/) - [Donald Metzler](https://md.research.elamur.ai/free/en/authors/donald-metzler/) - [Huaixiu Steven Zheng](https://md.research.elamur.ai/free/en/authors/huaixiu-steven-zheng/) - [Hyung Won Chung](https://md.research.elamur.ai/free/en/authors/hyung-won-chung/) - [Jamie Hall](https://md.research.elamur.ai/free/en/authors/jamie-hall/) - [Jason Wei](https://md.research.elamur.ai/free/en/authors/jason-wei/) - [Jeff Dean](https://md.research.elamur.ai/free/en/authors/jeff-dean/) - [Jinfeng Rao](https://md.research.elamur.ai/free/en/authors/jinfeng-rao/) - [Joseph E. Gonzalez](https://md.research.elamur.ai/free/en/authors/joseph-e-gonzalez/) ## Cited by Papers of our library cite their works 1,371 times (of the 140,884 papers whose bibliographies we read); [141 citing papers with a page here →](https://md.research.elamur.ai/free/en/authors/david-r-so/cited-by/).