The company said that the model was trained on 15 trillion mixed visual and text tokens.
Back To Top
Subscribe to our weekly newsletter bellow and never miss the latest updates