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

Add to Chrome

Add to Firefox

Picture for Danish Pruthi

Danish Pruthi

Learning the Legibility of Visual Text Perturbations

Mar 10, 2023
Dev Seth, Rickard Stureborg, Danish Pruthi, Bhuwan Dhingra

Figure 1 for Learning the Legibility of Visual Text Perturbations

Figure 2 for Learning the Legibility of Visual Text Perturbations

Figure 3 for Learning the Legibility of Visual Text Perturbations

Figure 4 for Learning the Legibility of Visual Text Perturbations

* 14 pages, 7 figures. Accepted at EACL 2023 (main, long)

Access Paper or Ask Questions

Measures of Information Reflect Memorization Patterns

Oct 19, 2022
Rachit Bansal, Danish Pruthi, Yonatan Belinkov

Figure 1 for Measures of Information Reflect Memorization Patterns

Figure 2 for Measures of Information Reflect Memorization Patterns

Figure 3 for Measures of Information Reflect Memorization Patterns

Figure 4 for Measures of Information Reflect Memorization Patterns

* 22 pages; NeurIPS 2022. Code and data at https://linktr.ee/InformationMeasures

Access Paper or Ask Questions

Do Context-Aware Translation Models Pay the Right Attention?

May 21, 2021
Kayo Yin, Patrick Fernandes, Danish Pruthi, Aditi Chaudhary, André F. T. Martins, Graham Neubig

Figure 1 for Do Context-Aware Translation Models Pay the Right Attention?

Figure 2 for Do Context-Aware Translation Models Pay the Right Attention?

Figure 3 for Do Context-Aware Translation Models Pay the Right Attention?

Figure 4 for Do Context-Aware Translation Models Pay the Right Attention?

* Accepted to ACL 2021

Access Paper or Ask Questions

Weakly- and Semi-supervised Evidence Extraction

Nov 03, 2020
Danish Pruthi, Bhuwan Dhingra, Graham Neubig, Zachary C. Lipton

Figure 1 for Weakly- and Semi-supervised Evidence Extraction

Figure 2 for Weakly- and Semi-supervised Evidence Extraction

Figure 3 for Weakly- and Semi-supervised Evidence Extraction

Figure 4 for Weakly- and Semi-supervised Evidence Extraction

* Accepted to the Findings of EMNLP 2020, to be presented at BlackBoxNLP

Access Paper or Ask Questions