Strategy24
Results
2
comments of
Strategy24
As a hack you can make a constant input and pass it to a dense layer with softmax activation that will provide these probabilities.
I did like this `p = keras.layers.Dense(2, activation='softmax')(x)` and then `categorical = tensorflow_probability.distributions.Categorical(probs=tensorflow.reshape(p, (-1, 1, 2)))` for some reason `Categorical` needs tensor with extra dimension