Assuming that your (10,4) is a matrix which doesn't represent a 10 length sequence (where you will need an LSTM) OR an image (where you will need a 2D CNN), you can simply flatten() the input matrix and pass it through to the next few dense layers as below.
from tensorflow.keras import layers, Model
inp = layers.Input((10,4)) #none,10,4
x = layers.Flatten()(inp) #none,40
x = layers.Dense(256)(x) #none,256
out = layers.Dense(3)(x) #none,3
model = Model(inp, out)
model.summary()
Layer (type) Output Shape Param #
=================================================================
input_43 (InputLayer) [(None, 10, 4)] 0
_________________________________________________________________
flatten_1 (Flatten) (None, 40) 0
_________________________________________________________________
dense_82 (Dense) (None, 256) 10496
_________________________________________________________________
dense_83 (Dense) (None, 3) 771
=================================================================
Total params: 11,267
Trainable params: 11,267
Non-trainable params: 0
I tried this, but it doesn't work for me for some reason.
data.shapeTensorShape([10, 4])x = tf.keras.layers.Flatten()(data)x = tf.keras.layers.Dense(256)(x)x = tf.keras.layers.Dense(3)(x)x.shapeTensorShape([10, 4])it does work, my bad. Apparently the Flatten() Layer doesn't do anything to the matrix because it expects the first dimension to be the batch size. If I use model.fit(data) it will internally add the batch size as a dimension(i think) and work as you described.