Search code examples
tensorflowmachine-learningkerasartificial-intelligence

What is the difference between the `dataset.batch` function and the `batch_size` parameter of the `model.fit` function?

What is the difference between batching your dataset with dataset.batch(batch_size) and batching your dataset with the batch_size parameter on the .fit the function of your model? do they have the same functionality or are they different?

Solution

  • Check the documentation for the parameter batch_size in fit:

    batch_size
    Integer or None. Number of samples per gradient update. If unspecified, batch_size will default to 32. Do not specify the batch_size if your data is in the form of datasets, generators, or keras.utils.Sequence instances (since they generate batches).

    So, if you are passing a dataset object for training, do not use the batch_size parameter, as that is only meant for the case where your X/Y values are NumPy arrays or TensorFlow tensors.