Multimodal Models

Multimodal Zero-Shot Learning

Ability of a model to perform a task on one modality (e.g., classifying an image) without having been explicitly trained for it, by leveraging knowledge transferred from another modality (e.g., the text of class labels).

← Zurück