Filters








10 Hits in 9.9 sec

Thinking Aloud: Dynamic Context Generation Improves Zero-Shot Reasoning Performance of GPT-2 [article]

Gregor Betz and Kyle Richardson and Christian Voigt
2021 arXiv   pre-print
Our main result is that dynamic problem elaboration significantly improves the zero-shot performance of GPT-2 in a deductive reasoning and natural language inference task: While the model uses a syntactic  ...  We suggest to improve the reasoning ability of pre-trained neural language models in a similar way, namely by expanding a task's context with problem elaborations that are dynamically generated by the  ...  Acknowledgements We would like to thank the members of the Aristo group at Allen AI for valuable feedback on earlier versions of this work.  ... 
arXiv:2103.13033v1 fatcat:svpsktwc35a2fcivy2ja4jvoti

PromptMaker: Prompt-based Prototyping with Large Language Models

Ellen Jiang, Kristen Olson, Edwin Toh, Alejandra Molina, Aaron Donsbach, Michael Terry, Carrie J Cai
2022 CHI Conference on Human Factors in Computing Systems Extended Abstracts  
Taken together, this case study provides important implications that lay the groundwork toward a new future of prototyping with ML.  ...  Prototyping is notoriously difficult to do with machine learning (ML), but recent advances in large language models may lower the barriers to people prototyping with ML, through the use of natural language  ...  Carrie Cai guided the research, conducted user studies and analysis, gave high-level scientific advice, and wrote much of the paper.  ... 
doi:10.1145/3491101.3503564 fatcat:mjxonbjkvnhi5b3lxq26uc2wsi

The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence [article]

Gary Marcus
2020 arXiv   pre-print
Recent research in artificial intelligence and machine learning has largely emphasized general-purpose learning and ever-larger training sets and more and more compute.  ...  In contrast, I propose a hybrid, knowledge-driven, reasoning-based approach, centered around cognitive models, that could provide the substrate for a richer, more robust AI than is currently possible.  ...  One can imagine improving a system like GPT-2 by gathering larger data sets, augmenting those data sets in various ways, and incorporating various kinds of improvements in the underlying architecture.  ... 
arXiv:2002.06177v3 fatcat:evtkborprvcbjmk5b4vznuklli

Go Back in Time: Generating Flashbacks in Stories with Event Temporal Prompts [article]

Rujun Han, Hong Chen, Yufei Tian, Nanyun Peng
2022 arXiv   pre-print
However, it is challenging for machines to generate flashback as it requires a solid understanding of event temporal order (e.g.  ...  Two major issues in existing systems that exacerbate the challenges: 1) temporal bias in pertaining and story datasets that leads to monotonic event temporal orders; 2) lack of explicit guidance that helps  ...  PPL: generated stories' perplexity scored by GPT-2 (Radford et al., 2019) , i.e. we feed the generated stories into GPT-2 to compute perplexity scores.  ... 
arXiv:2205.01898v1 fatcat:lcnaasa36rfbngbyf73viubgwa

Opal: Multimodal Image Generation for News Illustration [article]

Vivian Liu, Han Qiao, Lydia Chilton
2022 arXiv   pre-print
Our evaluation shows that Opal efficiently generates diverse sets of editorial illustrations, graphic assets, and concept ideas.  ...  Recent work has shown that text-to-image generations are able to represent a broad range of subjects and artistic styles. However, translating text prompts into visual messages is difficult.  ...  Two of the editorial illustrators in their think aloud process mentioned that they had a concept in mind at the start.  ... 
arXiv:2204.09007v2 fatcat:owtkmwq4rzfc5l7i7mxmhk5zeu

Machine translation for everyone: Empowering users in the age of artificial intelligence [article]

Dorothy Kenny
2022 Zenodo  
It also shows how users can make the most of MT through pre-editing, post-editing and customization of the technology.  ...  Both have been affected by the increasing availability of machine translation (MT): language learners now make use of free online MT to help them both understand and produce texts in a second language,  ...  Thanks also to Reinhard Rapp and Felix Kopecky for their thorough review of this chapter. Any remaining errors are mine alone.  ... 
doi:10.5281/zenodo.6653405 fatcat:fqpka6rnvrcqbpmchjt4u7wiwq

LaMPost: Design and Evaluation of an AI-assisted Email Writing Prototype for Adults with Dyslexia [article]

Steven M. Goodman, Erin Buehler, Patrick Clary, Andy Coenen, Aaron Donsbach, Tiffanie N. Horne, Michal Lahav, Robert Macdonald, Rain Breaw Michaels, Ajit Narayanan, Mahima Pushkarna, Joel Riley (+6 others)
2022 pre-print
the current generation of LLMs may not surpass the accuracy and quality thresholds required to meet the needs of writers with dyslexia.  ...  However, the capabilities for natural language generation demonstrated by the latest class of large language models (LLMs) highlight an opportunity to explore new forms of human-AI writing support tools  ...  Through a lens of neurodiversity, however, dyslexia is defned by heightened spatial and perceptual abilities, interconnected and dynamic reasoning, and narrative and holistic thinking-alongside commonly  ... 
doi:10.1145/3517428.3544819 arXiv:2207.02308v1 fatcat:ir2nem5etfe5tag47rhikrsom4

Go Back in Time: Generating Flashbacks in Stories with Event Temporal Prompts

Rujun Han, Hong Chen, Yufei Tian, Nanyun Peng
2022 Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies   unpublished
Two major issues in existing systems exacerbate the challenges: 1) temporal bias in pretraining and story datasets that leads to monotonic event temporal orders; 2) lack of explicit guidance that helps  ...  Stories or narratives are comprised of a sequence of events.  ...  PPL: generated stories' perplexity scored by GPT-2 (Radford et al., 2019) , i.e. we feed the generated stories into GPT-2 to compute perplexity scores.  ... 
doi:10.18653/v1/2022.naacl-main.104 fatcat:krafqmboevgb3gehdrip3y6jsa

Abstracts of the 60th Conference of Experimental Psychologists [article]

(:Unkn) Unknown, Leibniz Institut Für Psychologische Information Und Dokumentation (ZPID), Leibniz Institut Für Psychologische Information Und Dokumentation (ZPID), Alexander C. Schütz, Anna Schubö, Dominik Endres, Harald Lachnit
2018
task) or transient dynamics of cognitive control (a context-updating task).  ...  As a remedy, we propose Generalized Processing Tree (GPT) models as a flexible approach for the analysis of mouse-tracking data.  ...  Such adjustment is often inferred from modulations of congruency effects, that is, from variations in the size of the performance difference when target and distractor stimuli are associated with same  ... 
doi:10.23668/psycharchives.913 fatcat:thnbia5o4bfhxiuaejlfghduku

A B A Letter from the Editor Special anks

Faculty Advisor, Mark Brown, Graduate Advisor, Kate Wilkins, Keaton Sloan, Operations Associates, Victoria Crowe, Elizabeth Linxwiler, Nicholas Mcglashen, Tysha Medeiros, Erin Oppenheim, Quintn Parker (+45 others)
2014 Journal of Undergraduate Research and Scholarly Excellence   unpublished
We embarked on a complete rebranding of our organization, increased our social media presence fourfold, improved our community service and involvement, and expanded internationally.  ...  JUR Press is proud to present Volume V of the Journal of Undergraduate Research and Scholarly Excellence.  ...  The research for this article decodes tensions of the inner and outer lives of monks unique to the forest tradition.  ... 
fatcat:rqbg7xewsnahncjjgbnylrtbte