Output 层的形状始终为 (1,)

Question

I am trying to make a simple Keras model. But regardless of what output shape I am specifying the output layer is always of the shape (1,) so I cannot train my model because of the output layer and target data shapes mismatch.

import keras
from keras.models import Sequential
from keras.layers import InputLayer, LSTM, Dense

# 63 is the number of unique characters
# 128 is the length of a sequence of characters

X = ... # X is an one-hot ndarray; X.shape == (96092, 128, 63)
Y = ... # Y is an one-hot ndarray; Y.shape == (96092, 63)

model = Sequential()
model.add(InputLayer([128, 63]))
model.add(LSTM(96))
model.add(Dense(63))

model.compile(
  optimizer=keras.optimizers.RMSprop(1e-3, decay=1e-5),
  loss=keras.losses.sparse_categorical_crossentropy,
)

model.fit(X, Y) # ValueError: Error when checking target: expected dense_4 to have shape (1,) but got array with shape (63,)

As you can see, the shape of the output dense layer is (1,) but it has to be of shape (63,) . What am I doing wrong?

I am using Google Colab with preinstalled Keras.

Answer 1

What that error is saying is that the shape of the output layer is 63. However, for some reason, it expects 1.

In this case, the reason it expects 1 is because you are using sparse_categorical_crossentropy , which expects an integer representing the index of the output category. Instead, use categorical_crossentropy , which expects a one-hot encoding of the output category.