--- title: "Jason Yosinski — author profile and papers" description: "Jason Yosinski: 2 papers on Elamur Research (2019–2021), h-index 26 — affiliations, topics, co-authors and who cites them in our library." canonical: "https://research.elamur.ai/free/en/authors/jason-yosinski/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/authors/jason-yosinski/) [home](https://md.research.elamur.ai/free/en/) / [authors](https://md.research.elamur.ai/free/en/authors/) / Jason Yosinski # Jason Yosinski Jason Yosinski's papers here deal with large pretrained language models and how to get useful behaviour out of them without retraining. One line of work looks at steering generated text toward a chosen topic or sentiment. Another asks how far models trained mostly on English can handle tasks in other languages when given only a few examples in the prompt. In both, the model's weights stay fixed, and the work is done by small add-on components or carefully built prompts, tested on models such as GPT-2. The most cited of these papers is [Plug and Play Language Models](https://md.research.elamur.ai/free/en/papers/dathathri-2019-plug-play-language-models-simple/) (2019), which introduced PPLM. It pairs a frozen, unconditional GPT-2 with lightweight attribute models, either a hand-chosen bag of words or a one-layer classifier on the model's hidden states. During decoding, gradients from that attribute model adjust the cached activations so the output drifts toward the target attribute, while extra terms keep the text close to the original distribution and therefore fluent. The work falls under [Text generation](https://md.research.elamur.ai/free/en/topics/text-generation/) and [Language models & pretraining](https://md.research.elamur.ai/free/en/topics/language-models/). The more recent direction turns to cross-lingual in-context learning. [Language Models are Few-shot Multilingual Learners](https://md.research.elamur.ai/free/en/papers/winata-2021-language-models-are-few-shot/) (2021) tests GPT-2, GPT NEO, GPT NEO-J and T5 on intent classification across SNIPS, MTOP and MultiNLU. It recasts each intent label as its own true-or-false question with a short dedicated prompt, then compares the normalised probability of "true" across labels. It also probes cross-lingual ability by placing English examples in the context while the query is in another language. This places the work within [Multilingual & cross-lingual NLP](https://md.research.elamur.ai/free/en/topics/multilingual-nlp/). Earlier bylines list [Cornell University](https://md.research.elamur.ai/free/en/universities/cornell-university/) (2011–2018) and Uber AI (2017–2020). Co-authors on these papers include [Sumanth Dathathri](https://md.research.elamur.ai/free/en/authors/sumanth-dathathri/), [Rosanne Liu](https://md.research.elamur.ai/free/en/authors/rosanne-liu/), [Andrea Madotto](https://md.research.elamur.ai/free/en/authors/andrea-madotto/), [Genta Indra Winata](https://md.research.elamur.ai/free/en/authors/genta-indra-winata/), Pascale Fung and Zhaojiang Lin. Jason Yosinski has 2 papers in our library, in 2019–2021, most of them on [Language models & pretraining](https://md.research.elamur.ai/free/en/topics/language-models/), [Text generation](https://md.research.elamur.ai/free/en/topics/text-generation/), [Multilingual & cross-lingual NLP](https://md.research.elamur.ai/free/en/topics/multilingual-nlp/); often with [Andrea Madotto](https://md.research.elamur.ai/free/en/authors/andrea-madotto/), [Rosanne Liu](https://md.research.elamur.ai/free/en/authors/rosanne-liu/), [Genta Indra Winata](https://md.research.elamur.ai/free/en/authors/genta-indra-winata/). 2 papers in our library · 2019–2021 [OpenAlex](https://openalex.org/A5049659753) · [**h-index 26**](https://openalex.org/authors/A5049659753) · [61 works](https://openalex.org/works?filter=author.id:A5049659753) · [12,193 citations](https://openalex.org/works?filter=author.id:A5049659753&sort=cited_by_count:desc) OpenAlex, 2026-10-14 [📄 2 works in our library](https://md.research.elamur.ai/free/en/authors/jason-yosinski/#papers) · [🔗 1,738 citations in our library](https://md.research.elamur.ai/free/en/authors/jason-yosinski/cited-by/) · [1 more in our library without a page yet](#also) · [⭐ Paper Star rank #2402](https://md.research.elamur.ai/free/en/paper-stars/page/25/#star-2402) 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=Jason+Yosinski) · [Semantic Scholar](https://www.semanticscholar.org/search?q=Jason+Yosinski) · [dblp](https://dblp.org/search?q=Jason+Yosinski) · [ACL Anthology](https://aclanthology.org/search/?q=Jason+Yosinski) ## Affiliations [🎓**Cornell University** Ithaca · United States 2011–2018 · 24 papers](https://md.research.elamur.ai/free/en/universities/cornell-university/) · [🏢**Uber AI** San Francisco · United States 2017–2020 · 7 papers](https://ror.org/05vm0ed18) · [🏢**Numerica Corporation** Fort Collins · United States 2008–2013 · 6 papers](https://ror.org/01j5et233) From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 5 more institutions on single bylines are not listed. ## Topics [Language models & pretraining (2)](https://md.research.elamur.ai/free/en/topics/language-models/) · [Text generation (1)](https://md.research.elamur.ai/free/en/topics/text-generation/) · [Multilingual & cross-lingual NLP (1)](https://md.research.elamur.ai/free/en/topics/multilingual-nlp/) ## Works by year ### 2020s [**2021** 1 paper](https://md.research.elamur.ai/free/en/papers/winata-2021-language-models-are-few-shot/) ### 2010s [**2019** 1 paper](https://md.research.elamur.ai/free/en/papers/dathathri-2019-plug-play-language-models-simple/) ## Papers The most cited first (OpenAlex and our library). 1. **[Plug and Play Language Models: A Simple Approach to Controlled Text Generation](https://md.research.elamur.ai/free/en/papers/dathathri-2019-plug-play-language-models-simple/)** *(2019 · 562 citations)* Large pretrained Transformer language models write fluent text, but steering what they produce, such as the topic or sentiment, usually means changing the architecture or retraining on labelled data. 2. **[Language Models are Few-shot Multilingual Learners](https://md.research.elamur.ai/free/en/papers/winata-2021-language-models-are-few-shot/)** *(2021 · 173 citations)* The paper asks whether large language models trained mostly on English can perform multi-class intent classification in other languages through in-context learning alone, with no parameter updates. ## Also in our library 1 more work of theirs is in our library without a page here yet; it will appear as we publish them. ## Co-authors here 6 co-authors with a page, the most papers together first. - [Andrea Madotto](https://md.research.elamur.ai/free/en/authors/andrea-madotto/) ×2 - [Rosanne Liu](https://md.research.elamur.ai/free/en/authors/rosanne-liu/) ×2 - [Genta Indra Winata](https://md.research.elamur.ai/free/en/authors/genta-indra-winata/) - [Pascale Fung](https://md.research.elamur.ai/free/en/authors/pascale-fung/) - [Sumanth Dathathri](https://md.research.elamur.ai/free/en/authors/sumanth-dathathri/) - [Zhaojiang Lin](https://md.research.elamur.ai/free/en/authors/zhaojiang-lin/) ## Cited by Papers of our library cite their works 1,738 times (of the 140,884 papers whose bibliographies we read); [177 citing papers with a page here →](https://md.research.elamur.ai/free/en/authors/jason-yosinski/cited-by/).