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

Add to Chrome

Add to Firefox

Picture for Roberto Dessì

Roberto Dessì

Can discrete information extraction prompts generalize across language models?

Mar 07, 2023
Nathanaël Carraz Rakotonirina, Roberto Dessì, Fabio Petroni, Sebastian Riedel, Marco Baroni

Figure 1 for Can discrete information extraction prompts generalize across language models?

Figure 2 for Can discrete information extraction prompts generalize across language models?

Figure 3 for Can discrete information extraction prompts generalize across language models?

Figure 4 for Can discrete information extraction prompts generalize across language models?

* Published as conference paper at ICLR 2023

Access Paper or Ask Questions

Interpretable agent communication from scratch(with a generic visual processor emerging on the side)

Jun 08, 2021
Roberto Dessì, Eugene Kharitonov, Marco Baroni

Figure 1 for Interpretable agent communication from scratch(with a generic visual processor emerging on the side)

Figure 2 for Interpretable agent communication from scratch(with a generic visual processor emerging on the side)

Figure 3 for Interpretable agent communication from scratch(with a generic visual processor emerging on the side)

Figure 4 for Interpretable agent communication from scratch(with a generic visual processor emerging on the side)

* 9 pages main text, 13 pages total

Access Paper or Ask Questions

Focus on What's Informative and Ignore What's not: Communication Strategies in a Referential Game

Nov 05, 2019
Roberto Dessì, Diane Bouchacourt, Davide Crepaldi, Marco Baroni

Figure 1 for Focus on What's Informative and Ignore What's not: Communication Strategies in a Referential Game

* 3rd NeurIPS Workshop on Emergent Communication

Access Paper or Ask Questions

CNNs found to jump around more skillfully than RNNs: Compositional generalization in seq2seq convolutional networks

May 21, 2019
Roberto Dessì, Marco Baroni

Figure 1 for CNNs found to jump around more skillfully than RNNs: Compositional generalization in seq2seq convolutional networks

Figure 2 for CNNs found to jump around more skillfully than RNNs: Compositional generalization in seq2seq convolutional networks

Figure 3 for CNNs found to jump around more skillfully than RNNs: Compositional generalization in seq2seq convolutional networks

Figure 4 for CNNs found to jump around more skillfully than RNNs: Compositional generalization in seq2seq convolutional networks

* accepted as a short paper at ACL 2019

Access Paper or Ask Questions