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

Add to Chrome

Add to Firefox

Picture for Roger Levy

Roger Levy

Analyzing Wrap-Up Effects through an Information-Theoretic Lens

Mar 31, 2022
Clara Meister, Tiago Pimentel, Thomas Hikaru Clark, Ryan Cotterell, Roger Levy

Figure 1 for Analyzing Wrap-Up Effects through an Information-Theoretic Lens

Figure 2 for Analyzing Wrap-Up Effects through an Information-Theoretic Lens

Figure 3 for Analyzing Wrap-Up Effects through an Information-Theoretic Lens

Figure 4 for Analyzing Wrap-Up Effects through an Information-Theoretic Lens

* ACL 2022 (main conference)

Access Paper or Ask Questions

Revisiting the Uniform Information Density Hypothesis

Sep 23, 2021
Clara Meister, Tiago Pimentel, Patrick Haller, Lena Jäger, Ryan Cotterell, Roger Levy

Figure 1 for Revisiting the Uniform Information Density Hypothesis

Figure 2 for Revisiting the Uniform Information Density Hypothesis

Figure 3 for Revisiting the Uniform Information Density Hypothesis

Figure 4 for Revisiting the Uniform Information Density Hypothesis

* Proceedings of EMNLP 2021

Access Paper or Ask Questions

Controlled Evaluation of Grammatical Knowledge in Mandarin Chinese Language Models

Sep 22, 2021
Yiwen Wang, Jennifer Hu, Roger Levy, Peng Qian

Figure 1 for Controlled Evaluation of Grammatical Knowledge in Mandarin Chinese Language Models

Figure 2 for Controlled Evaluation of Grammatical Knowledge in Mandarin Chinese Language Models

Figure 3 for Controlled Evaluation of Grammatical Knowledge in Mandarin Chinese Language Models

Figure 4 for Controlled Evaluation of Grammatical Knowledge in Mandarin Chinese Language Models

* To appear in the Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing (EMNLP 2021)

Access Paper or Ask Questions

Scalable pragmatic communication via self-supervision

Aug 12, 2021
Jennifer Hu, Roger Levy, Noga Zaslavsky

Figure 1 for Scalable pragmatic communication via self-supervision

Figure 2 for Scalable pragmatic communication via self-supervision

Figure 3 for Scalable pragmatic communication via self-supervision

* Workshop on Self-Supervised Learning @ ICML 2021

Access Paper or Ask Questions

Structural Guidance for Transformer Language Models

Jul 30, 2021
Peng Qian, Tahira Naseem, Roger Levy, Ramón Fernandez Astudillo

Figure 1 for Structural Guidance for Transformer Language Models

Figure 2 for Structural Guidance for Transformer Language Models

Figure 3 for Structural Guidance for Transformer Language Models

Figure 4 for Structural Guidance for Transformer Language Models

* To be issued as paper revision for ACL 2021

Access Paper or Ask Questions