ANLI (Adversarial NLI)

A three-round adversarial natural language inference benchmark where annotators iteratively wrote examples to fool the strongest model trained on all prior rounds.

Also known as: Adversarial NLI

unassessed

This page is a discovery lead. Nobody has yet assessed it against the catalogue contract, so it carries no disposition. Absence of evidence here is not evidence of staleness.
Categoryreasoning
Subcategoryadversarial natural language inference
Page statusunknown
Metricaccuracy
Directionhigher_is_better
Unit%
Dataset size169265
Dataset licenceCC BY-NC 4.0
PublisherFacebook AI Research (FAIR), with the University of North Carolina at Chapel Hill

What it measures

ANLI tests natural language inference: given a short context passage and a one-sentence hypothesis, a model must decide whether the hypothesis is entailed by the context, contradicted by it, or neither (neutral). What separates ANLI from earlier inference datasets like SNLI and MNLI is not the task format but how its examples were built. Human annotators were shown a context and a target label, then wrote a hypothesis specifically intended to fool a state-of-the-art NLI model already trained on all prior data; an item only entered the dataset once it beat that model and passed a separate human-verification pass. Three rounds (R1, R2, R3) repeated this loop with progressively stronger target models and, from round three onward, a wider range of source genres beyond Wikipedia.

Task format

Three-way classification: given a context and a hypothesis, label the pair as entailment, contradiction, or neutral.

Models reporting this benchmark

No model card in ModelSpec reports this benchmark yet.

Data

This page as JSON · Edit on GitHub