Для приведенного ниже примера AutoKeras 1.0:
from tensorflow.keras.datasets import mnist
import autokeras as ak
(x_train, y_train), (x_test, y_test) = mnist.load_data()
clf = ak.ImageClassifier(max_trials=10) # It tries 10 different models.
clf.fit(x_train, y_train,epochs=3)
После тренировки вывод выглядит следующим образом:
Epoch 1/3
1500/1500 [==============================] - 254s 169ms/step - loss: 0.1751 - accuracy: 0.9463 - val_loss: 0.0634 - val_accuracy: 0.9818
Epoch 2/3
1500/1500 [==============================] - 259s 173ms/step - loss: 0.0779 - accuracy: 0.9756 - val_loss: 0.0534 - val_accuracy: 0.9843
Epoch 3/3
1500/1500 [==============================] - 220s 147ms/step - loss: 0.0624 - accuracy: 0.9800 - val_loss: 0.0470 - val_accuracy: 0.9872
[Trial complete]
[Trial summary]
|-Trial ID: e2ba122017b8f0d6e3c8eed7c6c208ab
|-Score: 0.04698652009871633
|-Best step: 0
> Hyperparameters:
|-classification_head_1/dropout_rate: 0.5
|-classification_head_1/spatial_reduction_1/reduction_type: flatten
|-dense_block_1/dropout_rate: 0
|-dense_block_1/num_layers: 1
|-dense_block_1/units_0: 128
|-dense_block_1/use_batchnorm: False
|-image_block_1/augment: False
|-image_block_1/block_type: vanilla
|-image_block_1/conv_block_1/dropout_rate: 0.25
|-image_block_1/conv_block_1/filters_0_0: 32
|-image_block_1/conv_block_1/filters_0_1: 64
|-image_block_1/conv_block_1/kernel_size: 3
|-image_block_1/conv_block_1/max_pooling: True
|-image_block_1/conv_block_1/num_blocks: 1
|-image_block_1/conv_block_1/num_layers: 2
|-image_block_1/conv_block_1/separable: False
|-image_block_1/normalize: True
|-optimizer: adam
Какова интерпретация содержания ниже "Сводка по пробным версиям"? Что представляет собой «Лучший шаг»?