Model Titles

[nisia97]

Hello,

I am new to autokeras and am running into an issue recreating a model that is determined by autokeras trials.

Layer (type) Output Shape Param #

input_1 (InputLayer) [(None, 3)] 0

---

multi_category_encoding (Mul (None, 3) 0

---

normalization (Normalization (None, 3) 7

---

dense (Dense) (None, 1024) 4096

---

re_lu (ReLU) (None, 1024) 0

---

dense_1 (Dense) (None, 64) 65600

---

re_lu_1 (ReLU) (None, 64) 0

---

dropout (Dropout) (None, 64) 0

---

regression_head_1 (Dense) (None, 1845) 119925

Total params: 189,628
Trainable params: 189,621
Non-trainable params: 7

---

Here, I am not sure specifically what kind of layer is supposed to be a multi_category_encoding layer or a Noemalization layer as these are not options when building a model with keras. I have tried to load the model and train it where it left off, but that does not work either. Any help is greatly appreciated!

[haifeng-jin]

AutoKeras use some custom layers https://github.com/keras-team/autokeras/blob/master/autokeras/keras_layers.py