ValueError: 层序列 33 的输入 0 与层不兼容：预期 ndim=3，发现 ndim=2。 收到的完整形状：[64, 100]

Question

I am following this guide to learn to build a Simple RNN. Different to the guide I just want my model to predict the next int in a ascending sequence (eg x = [1,2,3] y = [2,3,4]) But when attempting to train my model I receive this error message:

ValueError: Input 0 of layer sequential_33 is incompatible with the layer: expected ndim=3, found ndim=2. Full shape received: [64, 100]

Like in the guide my dataset has shape:

<BatchDataset shapes: ((64, 100), (64, 100)), types: (tf.int64, tf.int64)>

A little different form the guide my Model is defined as

BATCH_SIZE = 64
n_neurons  = 101
model = Sequential()
# shape [batch_size, timesteps, features]
model.add(Input(batch_input_shape = (BATCH_SIZE,100,1)))
model.add(LSTM(n_neurons ,return_sequences=True, stateful=True))
model.add(Dense(1))
model.compile(loss='mean_squared_error', optimizer='adam', metrics=['accuracy'])
print(model.summary())

with the summary beeing:

Model: "sequential_37"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
lstm_40 (LSTM)               (64, 100, 101)            41612     
_________________________________________________________________
dense_28 (Dense)             (64, 100, 1)              102       
=================================================================
Total params: 41,714
Trainable params: 41,714
Non-trainable params: 0
_________________________________________________________________
None

Could you help me understand why I get this error, and how to fix it?

I have made sure that the dataset has the same dimensions as in the guide, and provided the Input layer with the "batch_input_shape= (BATCH_SIZE,100,1)" because I learned that LSTMs need at least 3D data with shape [batch_size, timesteps, features]. So I am a confused where I'm still incorrect.

Any help yould be appreciated!

Answer 1

You should feed shape (64, 100, 1) to the model instead of (64, 100). Just add a dimension to your data