alexkoulakos/explain-then-predict

Source code for the BlackBoxNLP 2024 @ EMNLP paper "Enhancing adversarial robustness in Natural Language Inference using explanations"

21
/ 100
Experimental

This project helps Natural Language Processing (NLP) researchers and practitioners improve the reliability of their Natural Language Inference (NLI) models. It takes a pair of text sentences (a premise and a hypothesis) and evaluates if providing an intermediate explanation of their relationship makes the model more robust against subtly altered, misleading inputs. The output indicates whether the two sentences entail, contradict, or are neutral to each other, with increased confidence.

No commits in the last 6 months.

Use this if you are developing or evaluating NLI models and want to make them more resistant to adversarial attacks and deceptive text inputs.

Not ideal if you are looking for a general-purpose NLI model or a tool for tasks other than evaluating model robustness through explanations.

Natural Language Processing Model Robustness Adversarial Defense Textual Entailment AI Safety
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 5 / 25
Maturity 16 / 25
Community 0 / 25

How are scores calculated?

Stars

10

Forks

Language

Python

License

MIT

Last pushed

Nov 18, 2024

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/generative-ai/alexkoulakos/explain-then-predict"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.