joeddav/distilbert-base-uncased-agnews-student

The model is distilled from the zero-shot classification pipeline on the unlabeled AG's News dataset . It is primarily intended as a demo of how an expensive NLI-based model can be distilled to a more efficient student . The model will likely not perform as well as a model trained with full supervision . It can be used like any other model trained on AG's news, but will probably not perform well with supervision . For more information about the model, visit this notebook: "Teacher model: roberta-large-mnli," or "teacher hypothesis template: "This text is about {}" or "This is about a teacher," for example. For more details about,