Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox
Picture for Noah A. Smith

Noah A. Smith

Whose Language Counts as High Quality? Measuring Language Ideologies in Text Data Selection


Jan 25, 2022
Suchin Gururangan, Dallas Card, Sarah K. Drier, Emily K. Gade, Leroy Z. Wang, Zeyu Wang, Luke Zettlemoyer, Noah A. Smith

Figure 1 for Whose Language Counts as High Quality? Measuring Language Ideologies in Text Data Selection
Figure 2 for Whose Language Counts as High Quality? Measuring Language Ideologies in Text Data Selection
Figure 3 for Whose Language Counts as High Quality? Measuring Language Ideologies in Text Data Selection
Figure 4 for Whose Language Counts as High Quality? Measuring Language Ideologies in Text Data Selection

  Access Paper or Ask Questions

UnifiedSKG: Unifying and Multi-Tasking Structured Knowledge Grounding with Text-to-Text Language Models


Jan 20, 2022
Tianbao Xie, Chen Henry Wu, Peng Shi, Ruiqi Zhong, Torsten Scholak, Michihiro Yasunaga, Chien-Sheng Wu, Ming Zhong, Pengcheng Yin, Sida I. Wang, Victor Zhong, Bailin Wang, Chengzu Li, Connor Boyle, Ansong Ni, Ziyu Yao, Dragomir Radev, Caiming Xiong, Lingpeng Kong, Rui Zhang, Noah A. Smith, Luke Zettlemoyer, Tao Yu

Figure 1 for UnifiedSKG: Unifying and Multi-Tasking Structured Knowledge Grounding with Text-to-Text Language Models
Figure 2 for UnifiedSKG: Unifying and Multi-Tasking Structured Knowledge Grounding with Text-to-Text Language Models
Figure 3 for UnifiedSKG: Unifying and Multi-Tasking Structured Knowledge Grounding with Text-to-Text Language Models
Figure 4 for UnifiedSKG: Unifying and Multi-Tasking Structured Knowledge Grounding with Text-to-Text Language Models

  Access Paper or Ask Questions

WANLI: Worker and AI Collaboration for Natural Language Inference Dataset Creation


Jan 16, 2022
Alisa Liu, Swabha Swayamdipta, Noah A. Smith, Yejin Choi

Figure 1 for WANLI: Worker and AI Collaboration for Natural Language Inference Dataset Creation
Figure 2 for WANLI: Worker and AI Collaboration for Natural Language Inference Dataset Creation
Figure 3 for WANLI: Worker and AI Collaboration for Natural Language Inference Dataset Creation
Figure 4 for WANLI: Worker and AI Collaboration for Natural Language Inference Dataset Creation
* preprint 
  Access Paper or Ask Questions

Computational Lens on Cognition: Study Of Autobiographical Versus Imagined Stories With Large-Scale Language Models


Jan 07, 2022
Maarten Sap, Anna Jafarpour, Yejin Choi, Noah A. Smith, James W. Pennebaker, Eric Horvitz

Figure 1 for Computational Lens on Cognition: Study Of Autobiographical Versus Imagined Stories With Large-Scale Language Models
Figure 2 for Computational Lens on Cognition: Study Of Autobiographical Versus Imagined Stories With Large-Scale Language Models
Figure 3 for Computational Lens on Cognition: Study Of Autobiographical Versus Imagined Stories With Large-Scale Language Models
Figure 4 for Computational Lens on Cognition: Study Of Autobiographical Versus Imagined Stories With Large-Scale Language Models
* Equal contribution from Sap and Jafarpour; in review 
  Access Paper or Ask Questions

NeuroLogic A*esque Decoding: Constrained Text Generation with Lookahead Heuristics


Dec 16, 2021
Ximing Lu, Sean Welleck, Peter West, Liwei Jiang, Jungo Kasai, Daniel Khashabi, Ronan Le Bras, Lianhui Qin, Youngjae Yu, Rowan Zellers, Noah A. Smith, Yejin Choi

Figure 1 for NeuroLogic A*esque Decoding: Constrained Text Generation with Lookahead Heuristics
Figure 2 for NeuroLogic A*esque Decoding: Constrained Text Generation with Lookahead Heuristics
Figure 3 for NeuroLogic A*esque Decoding: Constrained Text Generation with Lookahead Heuristics
Figure 4 for NeuroLogic A*esque Decoding: Constrained Text Generation with Lookahead Heuristics

  Access Paper or Ask Questions

Bidimensional Leaderboards: Generate and Evaluate Language Hand in Hand


Dec 08, 2021
Jungo Kasai, Keisuke Sakaguchi, Ronan Le Bras, Lavinia Dunagan, Jacob Morrison, Alexander R. Fabbri, Yejin Choi, Noah A. Smith

Figure 1 for Bidimensional Leaderboards: Generate and Evaluate Language Hand in Hand
Figure 2 for Bidimensional Leaderboards: Generate and Evaluate Language Hand in Hand
Figure 3 for Bidimensional Leaderboards: Generate and Evaluate Language Hand in Hand
Figure 4 for Bidimensional Leaderboards: Generate and Evaluate Language Hand in Hand
* Project website: https://nlp.cs.washington.edu/billboard/ 
  Access Paper or Ask Questions

Transparent Human Evaluation for Image Captioning


Nov 17, 2021
Jungo Kasai, Keisuke Sakaguchi, Lavinia Dunagan, Jacob Morrison, Ronan Le Bras, Yejin Choi, Noah A. Smith

Figure 1 for Transparent Human Evaluation for Image Captioning
Figure 2 for Transparent Human Evaluation for Image Captioning
Figure 3 for Transparent Human Evaluation for Image Captioning
Figure 4 for Transparent Human Evaluation for Image Captioning

  Access Paper or Ask Questions

Annotators with Attitudes: How Annotator Beliefs And Identities Bias Toxic Language Detection


Nov 15, 2021
Maarten Sap, Swabha Swayamdipta, Laura Vianna, Xuhui Zhou, Yejin Choi, Noah A. Smith

Figure 1 for Annotators with Attitudes: How Annotator Beliefs And Identities Bias Toxic Language Detection
Figure 2 for Annotators with Attitudes: How Annotator Beliefs And Identities Bias Toxic Language Detection
Figure 3 for Annotators with Attitudes: How Annotator Beliefs And Identities Bias Toxic Language Detection
Figure 4 for Annotators with Attitudes: How Annotator Beliefs And Identities Bias Toxic Language Detection

  Access Paper or Ask Questions

Time Waits for No One! Analysis and Challenges of Temporal Misalignment


Nov 14, 2021
Kelvin Luu, Daniel Khashabi, Suchin Gururangan, Karishma Mandyam, Noah A. Smith

Figure 1 for Time Waits for No One! Analysis and Challenges of Temporal Misalignment
Figure 2 for Time Waits for No One! Analysis and Challenges of Temporal Misalignment
Figure 3 for Time Waits for No One! Analysis and Challenges of Temporal Misalignment
Figure 4 for Time Waits for No One! Analysis and Challenges of Temporal Misalignment
* 9 pages, 6 figures, 3 tables 
  Access Paper or Ask Questions

ABC: Attention with Bounded-memory Control


Oct 06, 2021
Hao Peng, Jungo Kasai, Nikolaos Pappas, Dani Yogatama, Zhaofeng Wu, Lingpeng Kong, Roy Schwartz, Noah A. Smith

Figure 1 for ABC: Attention with Bounded-memory Control
Figure 2 for ABC: Attention with Bounded-memory Control
Figure 3 for ABC: Attention with Bounded-memory Control
Figure 4 for ABC: Attention with Bounded-memory Control

  Access Paper or Ask Questions

Expected Validation Performance and Estimation of a Random Variable's Maximum


Oct 01, 2021
Jesse Dodge, Suchin Gururangan, Dallas Card, Roy Schwartz, Noah A. Smith

Figure 1 for Expected Validation Performance and Estimation of a Random Variable's Maximum
Figure 2 for Expected Validation Performance and Estimation of a Random Variable's Maximum
Figure 3 for Expected Validation Performance and Estimation of a Random Variable's Maximum
Figure 4 for Expected Validation Performance and Estimation of a Random Variable's Maximum

  Access Paper or Ask Questions

Sentence Bottleneck Autoencoders from Transformer Language Models


Aug 31, 2021
Ivan Montero, Nikolaos Pappas, Noah A. Smith

Figure 1 for Sentence Bottleneck Autoencoders from Transformer Language Models
Figure 2 for Sentence Bottleneck Autoencoders from Transformer Language Models
Figure 3 for Sentence Bottleneck Autoencoders from Transformer Language Models
Figure 4 for Sentence Bottleneck Autoencoders from Transformer Language Models

  Access Paper or Ask Questions

Train Short, Test Long: Attention with Linear Biases Enables Input Length Extrapolation


Aug 27, 2021
Ofir Press, Noah A. Smith, Mike Lewis

Figure 1 for Train Short, Test Long: Attention with Linear Biases Enables Input Length Extrapolation
Figure 2 for Train Short, Test Long: Attention with Linear Biases Enables Input Length Extrapolation
Figure 3 for Train Short, Test Long: Attention with Linear Biases Enables Input Length Extrapolation
Figure 4 for Train Short, Test Long: Attention with Linear Biases Enables Input Length Extrapolation

  Access Paper or Ask Questions

DEMix Layers: Disentangling Domains for Modular Language Modeling


Aug 20, 2021
Suchin Gururangan, Mike Lewis, Ari Holtzman, Noah A. Smith, Luke Zettlemoyer

Figure 1 for DEMix Layers: Disentangling Domains for Modular Language Modeling
Figure 2 for DEMix Layers: Disentangling Domains for Modular Language Modeling
Figure 3 for DEMix Layers: Disentangling Domains for Modular Language Modeling
Figure 4 for DEMix Layers: Disentangling Domains for Modular Language Modeling
* edits: updated reference links, added related work, typo fixes 
  Access Paper or Ask Questions

All That's 'Human' Is Not Gold: Evaluating Human Evaluation of Generated Text


Jul 07, 2021
Elizabeth Clark, Tal August, Sofia Serrano, Nikita Haduong, Suchin Gururangan, Noah A. Smith

Figure 1 for All That's 'Human' Is Not Gold: Evaluating Human Evaluation of Generated Text
Figure 2 for All That's 'Human' Is Not Gold: Evaluating Human Evaluation of Generated Text
Figure 3 for All That's 'Human' Is Not Gold: Evaluating Human Evaluation of Generated Text
Figure 4 for All That's 'Human' Is Not Gold: Evaluating Human Evaluation of Generated Text
* references added, corrected typo 
  Access Paper or Ask Questions

Scarecrow: A Framework for Scrutinizing Machine Text


Jul 06, 2021
Yao Dou, Maxwell Forbes, Rik Koncel-Kedziorski, Noah A. Smith, Yejin Choi

Figure 1 for Scarecrow: A Framework for Scrutinizing Machine Text
Figure 2 for Scarecrow: A Framework for Scrutinizing Machine Text
Figure 3 for Scarecrow: A Framework for Scrutinizing Machine Text
Figure 4 for Scarecrow: A Framework for Scrutinizing Machine Text
* The project webpage is at https://yao-dou.github.io/scarecrow/ 
  Access Paper or Ask Questions

On the Power of Saturated Transformers: A View from Circuit Complexity


Jun 30, 2021
William Merrill, Yoav Goldberg, Roy Schwartz, Noah A. Smith

Figure 1 for On the Power of Saturated Transformers: A View from Circuit Complexity
Figure 2 for On the Power of Saturated Transformers: A View from Circuit Complexity
Figure 3 for On the Power of Saturated Transformers: A View from Circuit Complexity
Figure 4 for On the Power of Saturated Transformers: A View from Circuit Complexity
* Preprint 
  Access Paper or Ask Questions

Specializing Multilingual Language Models: An Empirical Study


Jun 22, 2021
Ethan C. Chau, Noah A. Smith

Figure 1 for Specializing Multilingual Language Models: An Empirical Study
Figure 2 for Specializing Multilingual Language Models: An Empirical Study
Figure 3 for Specializing Multilingual Language Models: An Empirical Study
Figure 4 for Specializing Multilingual Language Models: An Empirical Study
* Work in progress 
  Access Paper or Ask Questions

Scientific Language Models for Biomedical Knowledge Base Completion: An Empirical Study


Jun 17, 2021
Rahul Nadkarni, David Wadden, Iz Beltagy, Noah A. Smith, Hannaneh Hajishirzi, Tom Hope

Figure 1 for Scientific Language Models for Biomedical Knowledge Base Completion: An Empirical Study
Figure 2 for Scientific Language Models for Biomedical Knowledge Base Completion: An Empirical Study
Figure 3 for Scientific Language Models for Biomedical Knowledge Base Completion: An Empirical Study
Figure 4 for Scientific Language Models for Biomedical Knowledge Base Completion: An Empirical Study

  Access Paper or Ask Questions

DExperts: Decoding-Time Controlled Text Generation with Experts and Anti-Experts


Jun 03, 2021
Alisa Liu, Maarten Sap, Ximing Lu, Swabha Swayamdipta, Chandra Bhagavatula, Noah A. Smith, Yejin Choi

Figure 1 for DExperts: Decoding-Time Controlled Text Generation with Experts and Anti-Experts
Figure 2 for DExperts: Decoding-Time Controlled Text Generation with Experts and Anti-Experts
Figure 3 for DExperts: Decoding-Time Controlled Text Generation with Experts and Anti-Experts
Figure 4 for DExperts: Decoding-Time Controlled Text Generation with Experts and Anti-Experts
* ACL 2021 camera-ready 
  Access Paper or Ask Questions

On-the-Fly Controlled Text Generation with Experts and Anti-Experts


May 07, 2021
Alisa Liu, Maarten Sap, Ximing Lu, Swabha Swayamdipta, Chandra Bhagavatula, Noah A. Smith, Yejin Choi

Figure 1 for On-the-Fly Controlled Text Generation with Experts and Anti-Experts
Figure 2 for On-the-Fly Controlled Text Generation with Experts and Anti-Experts
Figure 3 for On-the-Fly Controlled Text Generation with Experts and Anti-Experts
Figure 4 for On-the-Fly Controlled Text Generation with Experts and Anti-Experts
* Accepted to ACL 2021, camera-ready version coming soon 
  Access Paper or Ask Questions

A Dataset of Information-Seeking Questions and Answers Anchored in Research Papers


May 07, 2021
Pradeep Dasigi, Kyle Lo, Iz Beltagy, Arman Cohan, Noah A. Smith, Matt Gardner

Figure 1 for A Dataset of Information-Seeking Questions and Answers Anchored in Research Papers
Figure 2 for A Dataset of Information-Seeking Questions and Answers Anchored in Research Papers
Figure 3 for A Dataset of Information-Seeking Questions and Answers Anchored in Research Papers
Figure 4 for A Dataset of Information-Seeking Questions and Answers Anchored in Research Papers
* Accepted at NAACL 2021; Project page: https://allenai.org/project/qasper 
  Access Paper or Ask Questions

Provable Limitations of Acquiring Meaning from Ungrounded Form: What will Future Language Models Understand?


Apr 22, 2021
William Merrill, Yoav Goldberg, Roy Schwartz, Noah A. Smith

Figure 1 for Provable Limitations of Acquiring Meaning from Ungrounded Form: What will Future Language Models Understand?
Figure 2 for Provable Limitations of Acquiring Meaning from Ungrounded Form: What will Future Language Models Understand?
Figure 3 for Provable Limitations of Acquiring Meaning from Ungrounded Form: What will Future Language Models Understand?
Figure 4 for Provable Limitations of Acquiring Meaning from Ungrounded Form: What will Future Language Models Understand?
* Accepted at TACL; pre-MIT Press publication version 
  Access Paper or Ask Questions

Go Forth and Prosper: Language Modeling with Ancient Textual History


Apr 18, 2021
Rik Koncel-Kedziorski, Noah A. Smith

Figure 1 for Go Forth and Prosper: Language Modeling with Ancient Textual History
Figure 2 for Go Forth and Prosper: Language Modeling with Ancient Textual History

  Access Paper or Ask Questions

Probing Across Time: What Does RoBERTa Know and When?


Apr 16, 2021
Leo Z. Liu, Yizhong Wang, Jungo Kasai, Hannaneh Hajishirzi, Noah A. Smith

Figure 1 for Probing Across Time: What Does RoBERTa Know and When?
Figure 2 for Probing Across Time: What Does RoBERTa Know and When?
Figure 3 for Probing Across Time: What Does RoBERTa Know and When?
Figure 4 for Probing Across Time: What Does RoBERTa Know and When?

  Access Paper or Ask Questions