Natural Language Interface
Learn how to fine-tune the pre-trained BERT model for natural language interface (NLI) tasks.
We'll cover the following...
In natural language interface (NLI), the goal of our model is to determine whether a hypothesis is an entailment (true), a contradiction (false), or an undetermined (neutral) given a premise. Let's learn how to perform NLI by fine-tuning BERT.
Consider the sample dataset shown in the following figure; as we can see, we have a premise and a hypothesis with a label indicating whether they are entailment, contradiction, or undetermined:
Sample NLI dataset
Premise | Hypothesis | Label |
He is playing | He is sleeping | Contradiction |
A soccer game with multiple males playing | Some men are playing sports | Entailment |
An older and a younger man smiling | Two men are smiling at the dogs playing on the floor | Neutral |
Now, the goal of our model is to ...