Open rishabhrahulcom opened 8 months ago
Fixing it by adding a attribute _model in GenerateText() function
super().__init__()
self._model = None
This will leads to type cast issue in masked_loss() function
so casting it to int32.
def masked_loss(labels, preds): labels = tf.cast(labels, tf.int32)
GenerateText method in "image_captioning" in https://www.tensorflow.org/text/tutorials/image_captioning have no attribute model.