 What kind of tasks can transformers solve? 1. Translation Given the sentence in one language, transformers can generate sentences in another language. This is the traditional use case. 2. Image classification In 2020, Google introduced vision transformers that can determine the category an image belongs. 3. Speech recognition This involves the conversion of speech to text. WaveNet 2.0 is a transformer-based model that solves this problem. 4. Recommendation systems Given a sequence or a session of items viewed by a user, the next item or set of items can be recommended. NVIDIA's Transformers for Wreck is an open-source implementation of this concept. 5. Playing chess Given the state of the board, the transformer can determine the next best move. In fact, we can fine-tune a language model, like GPT-4, which is based on transformers, to learn this task.