Inicio Rápido

Introducción

En esta lección construimos nuestra primera red neuronal con Pytorch. Usamos el famoso ejemplo fashion mnist. En este caso tomaremos los datos directamente de la librería tensorflow.keras.datasets.

Instalar Tensorflow

En consola ejecute el siguiente comando.

#conda install -c conda-forge tensorflow

Para instalar sklearn use el canal de Intel, que mantiene la librería actualizada

# conda install -c intel scikit-learn

Carga librerías requeridas

import tensorflow as tf
from tensorflow.keras.models import Model
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Flatten, Dense
from tensorflow.keras.layers import Activation
from tensorflow.keras.layers import InputLayer

from tensorflow.keras.optimizers import SGD

from tensorflow.keras.datasets import cifar10

import matplotlib.pyplot as plt
import numpy as np

from sklearn.preprocessing import LabelBinarizer

2022-08-30 21:43:48.303939: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcudart.so.11.0'; dlerror: libcudart.so.11.0: cannot open shared object file: No such file or directory
2022-08-30 21:43:48.304002: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.

Trabajando con los datos

# Carga el dataset CIFAR-10 
print("Cargando el dataset CIFAR-10...")
((x_train, y_train), (x_test, y_test)) = cifar10.load_data()
print('¡¡Hecho!!')

assert x_train.shape == (50000, 32, 32, 3)
assert x_test.shape == (10000, 32, 32, 3)
assert y_train.shape == (50000, 1)
assert y_test.shape == (10000, 1)

# Escala los datos al rango [0, 1]
trainX = x_train.astype("float32") / 255.0
testX = x_test.astype("float32") / 255.0

# convierte las etiquetas de enteros a vectores (no necesario en realidad)
lb = LabelBinarizer()
trainY = lb.fit_transform(y_train)
testY = lb.transform(y_test)

rows = x_train.shape[1]
cols = x_train.shape[2]
channels = x_train.shape[3]

Cargando el dataset CIFAR-10...
Downloading data from https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz

     8192/170498071 [..............................] - ETA: 0s


    40960/170498071 [..............................] - ETA: 3:59


    90112/170498071 [..............................] - ETA: 3:37


   204800/170498071 [..............................] - ETA: 2:23


   385024/170498071 [..............................] - ETA: 1:41


   778240/170498071 [..............................] - ETA: 1:02


  1581056/170498071 [..............................] - ETA: 36s 


  3186688/170498071 [..............................] - ETA: 21s


  6316032/170498071 [>.............................] - ETA: 11s


  9412608/170498071 [>.............................] - ETA: 8s 


 12328960/170498071 [=>............................] - ETA: 7s


 15441920/170498071 [=>............................] - ETA: 6s


 18522112/170498071 [==>...........................] - ETA: 5s


 21651456/170498071 [==>...........................] - ETA: 5s


 24518656/170498071 [===>..........................] - ETA: 4s


 27648000/170498071 [===>..........................] - ETA: 4s


 30744576/170498071 [====>.........................] - ETA: 4s


 33873920/170498071 [====>.........................] - ETA: 3s


 36708352/170498071 [=====>........................] - ETA: 3s


 39788544/170498071 [======>.......................] - ETA: 3s


 42917888/170498071 [======>.......................] - ETA: 3s


 45801472/170498071 [=======>......................] - ETA: 3s


 48898048/170498071 [=======>......................] - ETA: 3s


 51863552/170498071 [========>.....................] - ETA: 3s


 54894592/170498071 [========>.....................] - ETA: 2s


 58040320/170498071 [=========>....................] - ETA: 2s


 61087744/170498071 [=========>....................] - ETA: 2s


 64217088/170498071 [==========>...................] - ETA: 2s


 67149824/170498071 [==========>...................] - ETA: 2s


 70230016/170498071 [===========>..................] - ETA: 2s


 73310208/170498071 [===========>..................] - ETA: 2s


 76095488/170498071 [============>.................] - ETA: 2s


 79175680/170498071 [============>.................] - ETA: 2s


 82305024/170498071 [=============>................] - ETA: 2s


 85155840/170498071 [=============>................] - ETA: 1s


 88285184/170498071 [==============>...............] - ETA: 1s


 91398144/170498071 [===============>..............] - ETA: 1s


 94199808/170498071 [===============>..............] - ETA: 1s


 97034240/170498071 [================>.............] - ETA: 1s


100016128/170498071 [================>.............] - ETA: 1s


102866944/170498071 [=================>............] - ETA: 1s


105947136/170498071 [=================>............] - ETA: 1s


108683264/170498071 [==================>...........] - ETA: 1s


111812608/170498071 [==================>...........] - ETA: 1s


114941952/170498071 [===================>..........] - ETA: 1s


118005760/170498071 [===================>..........] - ETA: 1s


120922112/170498071 [====================>.........] - ETA: 1s


124051456/170498071 [====================>.........] - ETA: 1s


125329408/170498071 [=====================>........] - ETA: 1s


128442368/170498071 [=====================>........] - ETA: 0s


130506752/170498071 [=====================>........] - ETA: 0s


133652480/170498071 [======================>.......] - ETA: 0s


136388608/170498071 [======================>.......] - ETA: 0s


139501568/170498071 [=======================>......] - ETA: 0s


142581760/170498071 [========================>.....] - ETA: 0s


145711104/170498071 [========================>.....] - ETA: 0s


148643840/170498071 [=========================>....] - ETA: 0s


151724032/170498071 [=========================>....] - ETA: 0s


154820608/170498071 [==========================>...] - ETA: 0s


157966336/170498071 [==========================>...] - ETA: 0s


160718848/170498071 [===========================>..] - ETA: 0s


163848192/170498071 [===========================>..] - ETA: 0s


166748160/170498071 [============================>.] - ETA: 0s


169861120/170498071 [============================>.] - ETA: 0s


170498071/170498071 [==============================] - 4s 0us/step

¡¡Hecho!!

print(len(trainY))
print(testY[0])

50000
[0 0 0 1 0 0 0 0 0 0]

Una primera imagen de los datos

# Muestra las primeras 100  images en color

img_rows = rows
img_cols = cols

imgs_t = x_test[:100]
imgs_t = imgs_t.reshape((10, 10, img_rows, img_cols, channels))
imgs_t = np.vstack([np.hstack(i) for i in imgs_t])
plt.figure(figsize=(8,8))
plt.axis('off')
plt.title('Primeras 100 imágenes de l conjunto de datos Cifar-10')
plt.imshow(imgs_t, interpolation='none')
plt.savefig('./test_color.png')
plt.show()

Creando modelos

# define model
class NeuralNetwork(Model):
    def __init__(self):
        super(NeuralNetwork, self).__init__()

        self.linear_relu_stack = Sequential([
            InputLayer(input_shape=(32,32,3)),
            Flatten(),
            Dense(512, activation='relu'),
            Dense(512, activation='relu'),
            Dense(10, activation='softmax'),            
        ])
                  
    # la función call es la que define la estructura de la red
    # en este ejemplo aceptamos solo una entrada, pero si lo desea,
    # siéntete libre de usar más
    def call(self, x):
        probs = self.linear_relu_stack(x)
        return probs

model = NeuralNetwork()

2022-08-30 21:43:59.859202: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcuda.so.1'; dlerror: libcuda.so.1: cannot open shared object file: No such file or directory
2022-08-30 21:43:59.859243: W tensorflow/stream_executor/cuda/cuda_driver.cc:269] failed call to cuInit: UNKNOWN ERROR (303)
2022-08-30 21:43:59.859265: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:156] kernel driver does not appear to be running on this host (fv-az449-310): /proc/driver/nvidia/version does not exist
2022-08-30 21:43:59.859548: I tensorflow/core/platform/cpu_feature_guard.cc:193] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  AVX2 AVX512F FMA
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.

Optimizador, métricas y compilación del modelo

init_lr = 1e-2
batch_size = 64
epochs = 10

optimizer = SGD(learning_rate=init_lr, momentum=0.9, decay=init_lr / epochs)

model.compile(loss="categorical_crossentropy", optimizer=optimizer,
              metrics=["accuracy"])

Entrenamiento

hist =  model.fit(trainX, trainY, batch_size=batch_size, 
                  validation_data=(testX, testY),
                  steps_per_epoch=trainX.shape[0] // batch_size,
                  epochs=epochs, verbose=1)

Epoch 1/10

2022-08-30 21:44:00.996372: W tensorflow/core/framework/cpu_allocator_impl.cc:82] Allocation of 614400000 exceeds 10% of free system memory.

  1/781 [..............................] - ETA: 7:10 - loss: 2.4179 - accuracy: 0.1250


  7/781 [..............................] - ETA: 7s - loss: 2.3998 - accuracy: 0.1071  


 14/781 [..............................] - ETA: 6s - loss: 2.3227 - accuracy: 0.1473


 21/781 [..............................] - ETA: 6s - loss: 2.2673 - accuracy: 0.1659


 28/781 [>.............................] - ETA: 6s - loss: 2.2260 - accuracy: 0.1808


 35/781 [>.............................] - ETA: 5s - loss: 2.1839 - accuracy: 0.2000


 42/781 [>.............................] - ETA: 5s - loss: 2.1758 - accuracy: 0.2016


 49/781 [>.............................] - ETA: 5s - loss: 2.1609 - accuracy: 0.2073


 56/781 [=>............................] - ETA: 5s - loss: 2.1451 - accuracy: 0.2121


 63/781 [=>............................] - ETA: 5s - loss: 2.1373 - accuracy: 0.2153


 70/781 [=>............................] - ETA: 5s - loss: 2.1253 - accuracy: 0.2208


 77/781 [=>............................] - ETA: 5s - loss: 2.1134 - accuracy: 0.2297


 84/781 [==>...........................] - ETA: 5s - loss: 2.1003 - accuracy: 0.2370


 91/781 [==>...........................] - ETA: 5s - loss: 2.0900 - accuracy: 0.2406


 98/781 [==>...........................] - ETA: 5s - loss: 2.0779 - accuracy: 0.2473


104/781 [==>...........................] - ETA: 5s - loss: 2.0652 - accuracy: 0.2509


111/781 [===>..........................] - ETA: 5s - loss: 2.0566 - accuracy: 0.2524


118/781 [===>..........................] - ETA: 5s - loss: 2.0467 - accuracy: 0.2572


125/781 [===>..........................] - ETA: 5s - loss: 2.0383 - accuracy: 0.2595


132/781 [====>.........................] - ETA: 5s - loss: 2.0254 - accuracy: 0.2631


139/781 [====>.........................] - ETA: 5s - loss: 2.0173 - accuracy: 0.2654


145/781 [====>.........................] - ETA: 5s - loss: 2.0127 - accuracy: 0.2678


152/781 [====>.........................] - ETA: 4s - loss: 2.0065 - accuracy: 0.2698


159/781 [=====>........................] - ETA: 4s - loss: 1.9997 - accuracy: 0.2731


166/781 [=====>........................] - ETA: 4s - loss: 1.9933 - accuracy: 0.2764


173/781 [=====>........................] - ETA: 4s - loss: 1.9873 - accuracy: 0.2790


180/781 [=====>........................] - ETA: 4s - loss: 1.9824 - accuracy: 0.2808


186/781 [======>.......................] - ETA: 4s - loss: 1.9802 - accuracy: 0.2816


193/781 [======>.......................] - ETA: 4s - loss: 1.9748 - accuracy: 0.2838


200/781 [======>.......................] - ETA: 4s - loss: 1.9695 - accuracy: 0.2853


207/781 [======>.......................] - ETA: 4s - loss: 1.9655 - accuracy: 0.2860


214/781 [=======>......................] - ETA: 4s - loss: 1.9615 - accuracy: 0.2875


221/781 [=======>......................] - ETA: 4s - loss: 1.9590 - accuracy: 0.2888


228/781 [=======>......................] - ETA: 4s - loss: 1.9552 - accuracy: 0.2900


234/781 [=======>......................] - ETA: 4s - loss: 1.9516 - accuracy: 0.2916


241/781 [========>.....................] - ETA: 4s - loss: 1.9480 - accuracy: 0.2927


248/781 [========>.....................] - ETA: 4s - loss: 1.9460 - accuracy: 0.2937


255/781 [========>.....................] - ETA: 4s - loss: 1.9418 - accuracy: 0.2958


261/781 [=========>....................] - ETA: 4s - loss: 1.9399 - accuracy: 0.2968


268/781 [=========>....................] - ETA: 4s - loss: 1.9369 - accuracy: 0.2977


275/781 [=========>....................] - ETA: 4s - loss: 1.9337 - accuracy: 0.2987


282/781 [=========>....................] - ETA: 3s - loss: 1.9293 - accuracy: 0.3005


288/781 [==========>...................] - ETA: 3s - loss: 1.9268 - accuracy: 0.3015


293/781 [==========>...................] - ETA: 3s - loss: 1.9240 - accuracy: 0.3029


300/781 [==========>...................] - ETA: 3s - loss: 1.9236 - accuracy: 0.3034


306/781 [==========>...................] - ETA: 3s - loss: 1.9238 - accuracy: 0.3041


313/781 [===========>..................] - ETA: 3s - loss: 1.9206 - accuracy: 0.3047


320/781 [===========>..................] - ETA: 3s - loss: 1.9200 - accuracy: 0.3052


327/781 [===========>..................] - ETA: 3s - loss: 1.9189 - accuracy: 0.3056


333/781 [===========>..................] - ETA: 3s - loss: 1.9178 - accuracy: 0.3062


340/781 [============>.................] - ETA: 3s - loss: 1.9152 - accuracy: 0.3071


346/781 [============>.................] - ETA: 3s - loss: 1.9154 - accuracy: 0.3067


352/781 [============>.................] - ETA: 3s - loss: 1.9136 - accuracy: 0.3074


359/781 [============>.................] - ETA: 3s - loss: 1.9105 - accuracy: 0.3089


365/781 [=============>................] - ETA: 3s - loss: 1.9080 - accuracy: 0.3093


371/781 [=============>................] - ETA: 3s - loss: 1.9051 - accuracy: 0.3106


377/781 [=============>................] - ETA: 3s - loss: 1.9031 - accuracy: 0.3115


383/781 [=============>................] - ETA: 3s - loss: 1.9012 - accuracy: 0.3122


390/781 [=============>................] - ETA: 3s - loss: 1.8993 - accuracy: 0.3130


396/781 [==============>...............] - ETA: 3s - loss: 1.8976 - accuracy: 0.3140


403/781 [==============>...............] - ETA: 3s - loss: 1.8938 - accuracy: 0.3154


409/781 [==============>...............] - ETA: 3s - loss: 1.8916 - accuracy: 0.3158


415/781 [==============>...............] - ETA: 2s - loss: 1.8904 - accuracy: 0.3164


421/781 [===============>..............] - ETA: 2s - loss: 1.8892 - accuracy: 0.3172


427/781 [===============>..............] - ETA: 2s - loss: 1.8881 - accuracy: 0.3179


434/781 [===============>..............] - ETA: 2s - loss: 1.8849 - accuracy: 0.3190


440/781 [===============>..............] - ETA: 2s - loss: 1.8835 - accuracy: 0.3194


444/781 [================>.............] - ETA: 2s - loss: 1.8819 - accuracy: 0.3199


449/781 [================>.............] - ETA: 2s - loss: 1.8816 - accuracy: 0.3201


455/781 [================>.............] - ETA: 2s - loss: 1.8793 - accuracy: 0.3211


461/781 [================>.............] - ETA: 2s - loss: 1.8774 - accuracy: 0.3220


468/781 [================>.............] - ETA: 2s - loss: 1.8753 - accuracy: 0.3230


473/781 [=================>............] - ETA: 2s - loss: 1.8736 - accuracy: 0.3237


479/781 [=================>............] - ETA: 2s - loss: 1.8725 - accuracy: 0.3243


486/781 [=================>............] - ETA: 2s - loss: 1.8708 - accuracy: 0.3251


493/781 [=================>............] - ETA: 2s - loss: 1.8680 - accuracy: 0.3261


499/781 [==================>...........] - ETA: 2s - loss: 1.8661 - accuracy: 0.3273


505/781 [==================>...........] - ETA: 2s - loss: 1.8645 - accuracy: 0.3281


511/781 [==================>...........] - ETA: 2s - loss: 1.8634 - accuracy: 0.3282


518/781 [==================>...........] - ETA: 2s - loss: 1.8622 - accuracy: 0.3290


524/781 [===================>..........] - ETA: 2s - loss: 1.8612 - accuracy: 0.3292


530/781 [===================>..........] - ETA: 2s - loss: 1.8596 - accuracy: 0.3296


536/781 [===================>..........] - ETA: 2s - loss: 1.8590 - accuracy: 0.3298


542/781 [===================>..........] - ETA: 1s - loss: 1.8581 - accuracy: 0.3301


548/781 [====================>.........] - ETA: 1s - loss: 1.8568 - accuracy: 0.3306


554/781 [====================>.........] - ETA: 1s - loss: 1.8559 - accuracy: 0.3311


560/781 [====================>.........] - ETA: 1s - loss: 1.8550 - accuracy: 0.3316


567/781 [====================>.........] - ETA: 1s - loss: 1.8529 - accuracy: 0.3326


573/781 [=====================>........] - ETA: 1s - loss: 1.8510 - accuracy: 0.3336


578/781 [=====================>........] - ETA: 1s - loss: 1.8498 - accuracy: 0.3340


584/781 [=====================>........] - ETA: 1s - loss: 1.8489 - accuracy: 0.3341


590/781 [=====================>........] - ETA: 1s - loss: 1.8474 - accuracy: 0.3348


596/781 [=====================>........] - ETA: 1s - loss: 1.8458 - accuracy: 0.3358


602/781 [======================>.......] - ETA: 1s - loss: 1.8443 - accuracy: 0.3362


608/781 [======================>.......] - ETA: 1s - loss: 1.8428 - accuracy: 0.3370


615/781 [======================>.......] - ETA: 1s - loss: 1.8408 - accuracy: 0.3377


622/781 [======================>.......] - ETA: 1s - loss: 1.8385 - accuracy: 0.3384


628/781 [=======================>......] - ETA: 1s - loss: 1.8373 - accuracy: 0.3392


635/781 [=======================>......] - ETA: 1s - loss: 1.8358 - accuracy: 0.3397


641/781 [=======================>......] - ETA: 1s - loss: 1.8353 - accuracy: 0.3399


648/781 [=======================>......] - ETA: 1s - loss: 1.8331 - accuracy: 0.3409


654/781 [========================>.....] - ETA: 1s - loss: 1.8321 - accuracy: 0.3416


661/781 [========================>.....] - ETA: 1s - loss: 1.8307 - accuracy: 0.3420


667/781 [========================>.....] - ETA: 0s - loss: 1.8294 - accuracy: 0.3425


674/781 [========================>.....] - ETA: 0s - loss: 1.8290 - accuracy: 0.3428


680/781 [=========================>....] - ETA: 0s - loss: 1.8282 - accuracy: 0.3434


687/781 [=========================>....] - ETA: 0s - loss: 1.8270 - accuracy: 0.3441


693/781 [=========================>....] - ETA: 0s - loss: 1.8259 - accuracy: 0.3444


700/781 [=========================>....] - ETA: 0s - loss: 1.8252 - accuracy: 0.3447


706/781 [==========================>...] - ETA: 0s - loss: 1.8240 - accuracy: 0.3451


712/781 [==========================>...] - ETA: 0s - loss: 1.8223 - accuracy: 0.3458


719/781 [==========================>...] - ETA: 0s - loss: 1.8202 - accuracy: 0.3466


725/781 [==========================>...] - ETA: 0s - loss: 1.8185 - accuracy: 0.3473


730/781 [===========================>..] - ETA: 0s - loss: 1.8175 - accuracy: 0.3476


736/781 [===========================>..] - ETA: 0s - loss: 1.8167 - accuracy: 0.3479


742/781 [===========================>..] - ETA: 0s - loss: 1.8155 - accuracy: 0.3487


748/781 [===========================>..] - ETA: 0s - loss: 1.8144 - accuracy: 0.3490


755/781 [============================>.] - ETA: 0s - loss: 1.8127 - accuracy: 0.3494


761/781 [============================>.] - ETA: 0s - loss: 1.8110 - accuracy: 0.3501


768/781 [============================>.] - ETA: 0s - loss: 1.8099 - accuracy: 0.3502


775/781 [============================>.] - ETA: 0s - loss: 1.8082 - accuracy: 0.3510


781/781 [==============================] - ETA: 0s - loss: 1.8069 - accuracy: 0.3515

2022-08-30 21:44:08.892887: W tensorflow/core/framework/cpu_allocator_impl.cc:82] Allocation of 122880000 exceeds 10% of free system memory.


781/781 [==============================] - 8s 10ms/step - loss: 1.8069 - accuracy: 0.3515 - val_loss: 1.6747 - val_accuracy: 0.4021

Epoch 2/10

  1/781 [..............................] - ETA: 6s - loss: 1.8293 - accuracy: 0.3125


  7/781 [..............................] - ETA: 6s - loss: 1.6688 - accuracy: 0.3425


 14/781 [..............................] - ETA: 6s - loss: 1.6655 - accuracy: 0.3785


 21/781 [..............................] - ETA: 6s - loss: 1.6715 - accuracy: 0.3966


 28/781 [>.............................] - ETA: 6s - loss: 1.6675 - accuracy: 0.3985


 35/781 [>.............................] - ETA: 6s - loss: 1.6591 - accuracy: 0.4042


 42/781 [>.............................] - ETA: 5s - loss: 1.6472 - accuracy: 0.4140


 47/781 [>.............................] - ETA: 6s - loss: 1.6453 - accuracy: 0.4166


 53/781 [=>............................] - ETA: 6s - loss: 1.6422 - accuracy: 0.4160


 60/781 [=>............................] - ETA: 6s - loss: 1.6406 - accuracy: 0.4146


 66/781 [=>............................] - ETA: 6s - loss: 1.6388 - accuracy: 0.4176


 72/781 [=>............................] - ETA: 6s - loss: 1.6408 - accuracy: 0.4171


 78/781 [=>............................] - ETA: 5s - loss: 1.6454 - accuracy: 0.4153


 84/781 [==>...........................] - ETA: 5s - loss: 1.6401 - accuracy: 0.4185


 91/781 [==>...........................] - ETA: 5s - loss: 1.6432 - accuracy: 0.4148


 98/781 [==>...........................] - ETA: 5s - loss: 1.6504 - accuracy: 0.4113


105/781 [===>..........................] - ETA: 5s - loss: 1.6493 - accuracy: 0.4116


111/781 [===>..........................] - ETA: 5s - loss: 1.6509 - accuracy: 0.4104


117/781 [===>..........................] - ETA: 5s - loss: 1.6478 - accuracy: 0.4126


124/781 [===>..........................] - ETA: 5s - loss: 1.6523 - accuracy: 0.4102


131/781 [====>.........................] - ETA: 5s - loss: 1.6530 - accuracy: 0.4093


138/781 [====>.........................] - ETA: 5s - loss: 1.6507 - accuracy: 0.4097


145/781 [====>.........................] - ETA: 5s - loss: 1.6500 - accuracy: 0.4096


152/781 [====>.........................] - ETA: 5s - loss: 1.6488 - accuracy: 0.4087


159/781 [=====>........................] - ETA: 5s - loss: 1.6484 - accuracy: 0.4097


166/781 [=====>........................] - ETA: 5s - loss: 1.6482 - accuracy: 0.4098


172/781 [=====>........................] - ETA: 5s - loss: 1.6443 - accuracy: 0.4115


178/781 [=====>........................] - ETA: 5s - loss: 1.6428 - accuracy: 0.4118


184/781 [======>.......................] - ETA: 4s - loss: 1.6434 - accuracy: 0.4116


190/781 [======>.......................] - ETA: 4s - loss: 1.6431 - accuracy: 0.4126


197/781 [======>.......................] - ETA: 4s - loss: 1.6413 - accuracy: 0.4132


203/781 [======>.......................] - ETA: 4s - loss: 1.6399 - accuracy: 0.4134


210/781 [=======>......................] - ETA: 4s - loss: 1.6397 - accuracy: 0.4140


216/781 [=======>......................] - ETA: 4s - loss: 1.6410 - accuracy: 0.4133


222/781 [=======>......................] - ETA: 4s - loss: 1.6417 - accuracy: 0.4133


228/781 [=======>......................] - ETA: 4s - loss: 1.6425 - accuracy: 0.4124


233/781 [=======>......................] - ETA: 4s - loss: 1.6422 - accuracy: 0.4123


237/781 [========>.....................] - ETA: 4s - loss: 1.6447 - accuracy: 0.4111


243/781 [========>.....................] - ETA: 4s - loss: 1.6459 - accuracy: 0.4115


249/781 [========>.....................] - ETA: 4s - loss: 1.6453 - accuracy: 0.4123


255/781 [========>.....................] - ETA: 4s - loss: 1.6447 - accuracy: 0.4126


261/781 [=========>....................] - ETA: 4s - loss: 1.6430 - accuracy: 0.4138


266/781 [=========>....................] - ETA: 4s - loss: 1.6427 - accuracy: 0.4138


273/781 [=========>....................] - ETA: 4s - loss: 1.6418 - accuracy: 0.4141


278/781 [=========>....................] - ETA: 4s - loss: 1.6415 - accuracy: 0.4137


284/781 [=========>....................] - ETA: 4s - loss: 1.6422 - accuracy: 0.4136


290/781 [==========>...................] - ETA: 4s - loss: 1.6415 - accuracy: 0.4136


295/781 [==========>...................] - ETA: 4s - loss: 1.6397 - accuracy: 0.4138


301/781 [==========>...................] - ETA: 4s - loss: 1.6388 - accuracy: 0.4140


307/781 [==========>...................] - ETA: 4s - loss: 1.6377 - accuracy: 0.4145


313/781 [===========>..................] - ETA: 4s - loss: 1.6371 - accuracy: 0.4145


319/781 [===========>..................] - ETA: 4s - loss: 1.6377 - accuracy: 0.4139


324/781 [===========>..................] - ETA: 3s - loss: 1.6386 - accuracy: 0.4138


330/781 [===========>..................] - ETA: 3s - loss: 1.6383 - accuracy: 0.4140


336/781 [===========>..................] - ETA: 3s - loss: 1.6375 - accuracy: 0.4144


342/781 [============>.................] - ETA: 3s - loss: 1.6371 - accuracy: 0.4144


348/781 [============>.................] - ETA: 3s - loss: 1.6370 - accuracy: 0.4143


354/781 [============>.................] - ETA: 3s - loss: 1.6359 - accuracy: 0.4145


361/781 [============>.................] - ETA: 3s - loss: 1.6362 - accuracy: 0.4147


368/781 [=============>................] - ETA: 3s - loss: 1.6364 - accuracy: 0.4150


374/781 [=============>................] - ETA: 3s - loss: 1.6351 - accuracy: 0.4156


380/781 [=============>................] - ETA: 3s - loss: 1.6346 - accuracy: 0.4162


386/781 [=============>................] - ETA: 3s - loss: 1.6332 - accuracy: 0.4165


392/781 [==============>...............] - ETA: 3s - loss: 1.6325 - accuracy: 0.4170


398/781 [==============>...............] - ETA: 3s - loss: 1.6329 - accuracy: 0.4170


404/781 [==============>...............] - ETA: 3s - loss: 1.6320 - accuracy: 0.4171


410/781 [==============>...............] - ETA: 3s - loss: 1.6319 - accuracy: 0.4171


417/781 [===============>..............] - ETA: 3s - loss: 1.6313 - accuracy: 0.4172


422/781 [===============>..............] - ETA: 3s - loss: 1.6302 - accuracy: 0.4180


427/781 [===============>..............] - ETA: 3s - loss: 1.6303 - accuracy: 0.4179


433/781 [===============>..............] - ETA: 3s - loss: 1.6309 - accuracy: 0.4175


439/781 [===============>..............] - ETA: 3s - loss: 1.6297 - accuracy: 0.4177


446/781 [================>.............] - ETA: 2s - loss: 1.6286 - accuracy: 0.4182


452/781 [================>.............] - ETA: 2s - loss: 1.6274 - accuracy: 0.4186


459/781 [================>.............] - ETA: 2s - loss: 1.6262 - accuracy: 0.4189


466/781 [================>.............] - ETA: 2s - loss: 1.6252 - accuracy: 0.4190


471/781 [=================>............] - ETA: 2s - loss: 1.6254 - accuracy: 0.4190


477/781 [=================>............] - ETA: 2s - loss: 1.6240 - accuracy: 0.4193


484/781 [=================>............] - ETA: 2s - loss: 1.6222 - accuracy: 0.4203


490/781 [=================>............] - ETA: 2s - loss: 1.6213 - accuracy: 0.4209


496/781 [==================>...........] - ETA: 2s - loss: 1.6192 - accuracy: 0.4217


502/781 [==================>...........] - ETA: 2s - loss: 1.6197 - accuracy: 0.4217


508/781 [==================>...........] - ETA: 2s - loss: 1.6185 - accuracy: 0.4225


514/781 [==================>...........] - ETA: 2s - loss: 1.6184 - accuracy: 0.4224


521/781 [===================>..........] - ETA: 2s - loss: 1.6179 - accuracy: 0.4227


528/781 [===================>..........] - ETA: 2s - loss: 1.6177 - accuracy: 0.4227


535/781 [===================>..........] - ETA: 2s - loss: 1.6162 - accuracy: 0.4233


541/781 [===================>..........] - ETA: 2s - loss: 1.6158 - accuracy: 0.4237


548/781 [====================>.........] - ETA: 2s - loss: 1.6142 - accuracy: 0.4243


555/781 [====================>.........] - ETA: 1s - loss: 1.6133 - accuracy: 0.4246


561/781 [====================>.........] - ETA: 1s - loss: 1.6136 - accuracy: 0.4245


567/781 [====================>.........] - ETA: 1s - loss: 1.6137 - accuracy: 0.4247


573/781 [=====================>........] - ETA: 1s - loss: 1.6128 - accuracy: 0.4248


580/781 [=====================>........] - ETA: 1s - loss: 1.6132 - accuracy: 0.4245


587/781 [=====================>........] - ETA: 1s - loss: 1.6131 - accuracy: 0.4247


594/781 [=====================>........] - ETA: 1s - loss: 1.6127 - accuracy: 0.4250


601/781 [======================>.......] - ETA: 1s - loss: 1.6119 - accuracy: 0.4250


608/781 [======================>.......] - ETA: 1s - loss: 1.6110 - accuracy: 0.4256


615/781 [======================>.......] - ETA: 1s - loss: 1.6095 - accuracy: 0.4261


622/781 [======================>.......] - ETA: 1s - loss: 1.6086 - accuracy: 0.4264


628/781 [=======================>......] - ETA: 1s - loss: 1.6088 - accuracy: 0.4266


635/781 [=======================>......] - ETA: 1s - loss: 1.6085 - accuracy: 0.4268


641/781 [=======================>......] - ETA: 1s - loss: 1.6083 - accuracy: 0.4267


648/781 [=======================>......] - ETA: 1s - loss: 1.6081 - accuracy: 0.4267


654/781 [========================>.....] - ETA: 1s - loss: 1.6076 - accuracy: 0.4267


661/781 [========================>.....] - ETA: 1s - loss: 1.6072 - accuracy: 0.4268


668/781 [========================>.....] - ETA: 0s - loss: 1.6076 - accuracy: 0.4271


674/781 [========================>.....] - ETA: 0s - loss: 1.6070 - accuracy: 0.4274


681/781 [=========================>....] - ETA: 0s - loss: 1.6065 - accuracy: 0.4280


688/781 [=========================>....] - ETA: 0s - loss: 1.6061 - accuracy: 0.4281


695/781 [=========================>....] - ETA: 0s - loss: 1.6054 - accuracy: 0.4284


702/781 [=========================>....] - ETA: 0s - loss: 1.6046 - accuracy: 0.4287


709/781 [==========================>...] - ETA: 0s - loss: 1.6047 - accuracy: 0.4285


716/781 [==========================>...] - ETA: 0s - loss: 1.6046 - accuracy: 0.4282


723/781 [==========================>...] - ETA: 0s - loss: 1.6039 - accuracy: 0.4284


730/781 [===========================>..] - ETA: 0s - loss: 1.6034 - accuracy: 0.4284


737/781 [===========================>..] - ETA: 0s - loss: 1.6033 - accuracy: 0.4282


743/781 [===========================>..] - ETA: 0s - loss: 1.6027 - accuracy: 0.4285


749/781 [===========================>..] - ETA: 0s - loss: 1.6020 - accuracy: 0.4289


755/781 [============================>.] - ETA: 0s - loss: 1.6012 - accuracy: 0.4291


761/781 [============================>.] - ETA: 0s - loss: 1.6005 - accuracy: 0.4293


767/781 [============================>.] - ETA: 0s - loss: 1.6002 - accuracy: 0.4296


773/781 [============================>.] - ETA: 0s - loss: 1.6001 - accuracy: 0.4297


780/781 [============================>.] - ETA: 0s - loss: 1.5996 - accuracy: 0.4298


781/781 [==============================] - 7s 9ms/step - loss: 1.5996 - accuracy: 0.4299 - val_loss: 1.6148 - val_accuracy: 0.4231

Epoch 3/10

  1/781 [..............................] - ETA: 7s - loss: 1.7400 - accuracy: 0.3906


  8/781 [..............................] - ETA: 6s - loss: 1.5740 - accuracy: 0.4440


 14/781 [..............................] - ETA: 6s - loss: 1.5496 - accuracy: 0.4469


 20/781 [..............................] - ETA: 6s - loss: 1.5259 - accuracy: 0.4489


 26/781 [..............................] - ETA: 6s - loss: 1.5255 - accuracy: 0.4517


 33/781 [>.............................] - ETA: 6s - loss: 1.5253 - accuracy: 0.4554


 40/781 [>.............................] - ETA: 6s - loss: 1.5194 - accuracy: 0.4578


 46/781 [>.............................] - ETA: 6s - loss: 1.5212 - accuracy: 0.4589


 52/781 [>.............................] - ETA: 6s - loss: 1.5109 - accuracy: 0.4640


 58/781 [=>............................] - ETA: 5s - loss: 1.5088 - accuracy: 0.4629


 64/781 [=>............................] - ETA: 5s - loss: 1.5086 - accuracy: 0.4644


 70/781 [=>............................] - ETA: 5s - loss: 1.5044 - accuracy: 0.4668


 76/781 [=>............................] - ETA: 5s - loss: 1.5070 - accuracy: 0.4649


 82/781 [==>...........................] - ETA: 5s - loss: 1.5060 - accuracy: 0.4637


 88/781 [==>...........................] - ETA: 5s - loss: 1.5125 - accuracy: 0.4615


 95/781 [==>...........................] - ETA: 5s - loss: 1.5142 - accuracy: 0.4627


102/781 [==>...........................] - ETA: 5s - loss: 1.5143 - accuracy: 0.4633


109/781 [===>..........................] - ETA: 5s - loss: 1.5142 - accuracy: 0.4632


116/781 [===>..........................] - ETA: 5s - loss: 1.5083 - accuracy: 0.4646


123/781 [===>..........................] - ETA: 5s - loss: 1.5085 - accuracy: 0.4668


130/781 [===>..........................] - ETA: 5s - loss: 1.5067 - accuracy: 0.4668


137/781 [====>.........................] - ETA: 5s - loss: 1.5127 - accuracy: 0.4649


144/781 [====>.........................] - ETA: 5s - loss: 1.5118 - accuracy: 0.4641


150/781 [====>.........................] - ETA: 5s - loss: 1.5148 - accuracy: 0.4618


157/781 [=====>........................] - ETA: 5s - loss: 1.5129 - accuracy: 0.4628


163/781 [=====>........................] - ETA: 5s - loss: 1.5100 - accuracy: 0.4644


170/781 [=====>........................] - ETA: 5s - loss: 1.5074 - accuracy: 0.4661


176/781 [=====>........................] - ETA: 4s - loss: 1.5083 - accuracy: 0.4659


183/781 [======>.......................] - ETA: 4s - loss: 1.5079 - accuracy: 0.4658


189/781 [======>.......................] - ETA: 4s - loss: 1.5081 - accuracy: 0.4641


195/781 [======>.......................] - ETA: 4s - loss: 1.5099 - accuracy: 0.4626


201/781 [======>.......................] - ETA: 4s - loss: 1.5135 - accuracy: 0.4623


207/781 [======>.......................] - ETA: 4s - loss: 1.5151 - accuracy: 0.4618


213/781 [=======>......................] - ETA: 4s - loss: 1.5142 - accuracy: 0.4616


219/781 [=======>......................] - ETA: 4s - loss: 1.5172 - accuracy: 0.4605


225/781 [=======>......................] - ETA: 4s - loss: 1.5174 - accuracy: 0.4607


231/781 [=======>......................] - ETA: 4s - loss: 1.5200 - accuracy: 0.4596


237/781 [========>.....................] - ETA: 4s - loss: 1.5203 - accuracy: 0.4589


242/781 [========>.....................] - ETA: 4s - loss: 1.5196 - accuracy: 0.4595


248/781 [========>.....................] - ETA: 4s - loss: 1.5214 - accuracy: 0.4578


254/781 [========>.....................] - ETA: 4s - loss: 1.5206 - accuracy: 0.4582


260/781 [========>.....................] - ETA: 4s - loss: 1.5219 - accuracy: 0.4570


265/781 [=========>....................] - ETA: 4s - loss: 1.5218 - accuracy: 0.4569


271/781 [=========>....................] - ETA: 4s - loss: 1.5207 - accuracy: 0.4577


277/781 [=========>....................] - ETA: 4s - loss: 1.5188 - accuracy: 0.4585


282/781 [=========>....................] - ETA: 4s - loss: 1.5169 - accuracy: 0.4594


288/781 [==========>...................] - ETA: 4s - loss: 1.5161 - accuracy: 0.4593


294/781 [==========>...................] - ETA: 4s - loss: 1.5168 - accuracy: 0.4590


300/781 [==========>...................] - ETA: 4s - loss: 1.5163 - accuracy: 0.4595


306/781 [==========>...................] - ETA: 4s - loss: 1.5156 - accuracy: 0.4599


312/781 [==========>...................] - ETA: 4s - loss: 1.5146 - accuracy: 0.4604


317/781 [===========>..................] - ETA: 4s - loss: 1.5152 - accuracy: 0.4603


323/781 [===========>..................] - ETA: 3s - loss: 1.5160 - accuracy: 0.4604


329/781 [===========>..................] - ETA: 3s - loss: 1.5159 - accuracy: 0.4608


335/781 [===========>..................] - ETA: 3s - loss: 1.5161 - accuracy: 0.4612


341/781 [============>.................] - ETA: 3s - loss: 1.5171 - accuracy: 0.4610


347/781 [============>.................] - ETA: 3s - loss: 1.5165 - accuracy: 0.4611


353/781 [============>.................] - ETA: 3s - loss: 1.5171 - accuracy: 0.4608


359/781 [============>.................] - ETA: 3s - loss: 1.5170 - accuracy: 0.4613


365/781 [=============>................] - ETA: 3s - loss: 1.5180 - accuracy: 0.4606


370/781 [=============>................] - ETA: 3s - loss: 1.5185 - accuracy: 0.4602


376/781 [=============>................] - ETA: 3s - loss: 1.5184 - accuracy: 0.4606


381/781 [=============>................] - ETA: 3s - loss: 1.5190 - accuracy: 0.4605


386/781 [=============>................] - ETA: 3s - loss: 1.5182 - accuracy: 0.4609


391/781 [==============>...............] - ETA: 3s - loss: 1.5179 - accuracy: 0.4610


397/781 [==============>...............] - ETA: 3s - loss: 1.5178 - accuracy: 0.4612


402/781 [==============>...............] - ETA: 3s - loss: 1.5179 - accuracy: 0.4610


408/781 [==============>...............] - ETA: 3s - loss: 1.5177 - accuracy: 0.4612


414/781 [==============>...............] - ETA: 3s - loss: 1.5177 - accuracy: 0.4611


420/781 [===============>..............] - ETA: 3s - loss: 1.5168 - accuracy: 0.4612


426/781 [===============>..............] - ETA: 3s - loss: 1.5162 - accuracy: 0.4612


432/781 [===============>..............] - ETA: 3s - loss: 1.5164 - accuracy: 0.4611


438/781 [===============>..............] - ETA: 3s - loss: 1.5171 - accuracy: 0.4606


444/781 [================>.............] - ETA: 2s - loss: 1.5166 - accuracy: 0.4608


450/781 [================>.............] - ETA: 2s - loss: 1.5171 - accuracy: 0.4607


457/781 [================>.............] - ETA: 2s - loss: 1.5178 - accuracy: 0.4605


463/781 [================>.............] - ETA: 2s - loss: 1.5173 - accuracy: 0.4609


469/781 [=================>............] - ETA: 2s - loss: 1.5170 - accuracy: 0.4611


475/781 [=================>............] - ETA: 2s - loss: 1.5167 - accuracy: 0.4616


482/781 [=================>............] - ETA: 2s - loss: 1.5166 - accuracy: 0.4620


488/781 [=================>............] - ETA: 2s - loss: 1.5162 - accuracy: 0.4624


495/781 [==================>...........] - ETA: 2s - loss: 1.5176 - accuracy: 0.4618


501/781 [==================>...........] - ETA: 2s - loss: 1.5180 - accuracy: 0.4615


507/781 [==================>...........] - ETA: 2s - loss: 1.5181 - accuracy: 0.4612


513/781 [==================>...........] - ETA: 2s - loss: 1.5174 - accuracy: 0.4614


519/781 [==================>...........] - ETA: 2s - loss: 1.5168 - accuracy: 0.4613


525/781 [===================>..........] - ETA: 2s - loss: 1.5173 - accuracy: 0.4610


531/781 [===================>..........] - ETA: 2s - loss: 1.5168 - accuracy: 0.4609


537/781 [===================>..........] - ETA: 2s - loss: 1.5168 - accuracy: 0.4605


543/781 [===================>..........] - ETA: 2s - loss: 1.5172 - accuracy: 0.4604


549/781 [====================>.........] - ETA: 2s - loss: 1.5172 - accuracy: 0.4605


555/781 [====================>.........] - ETA: 2s - loss: 1.5176 - accuracy: 0.4604


561/781 [====================>.........] - ETA: 1s - loss: 1.5183 - accuracy: 0.4603


567/781 [====================>.........] - ETA: 1s - loss: 1.5176 - accuracy: 0.4604


573/781 [=====================>........] - ETA: 1s - loss: 1.5174 - accuracy: 0.4607


579/781 [=====================>........] - ETA: 1s - loss: 1.5177 - accuracy: 0.4604


585/781 [=====================>........] - ETA: 1s - loss: 1.5167 - accuracy: 0.4608


592/781 [=====================>........] - ETA: 1s - loss: 1.5169 - accuracy: 0.4607


598/781 [=====================>........] - ETA: 1s - loss: 1.5165 - accuracy: 0.4608


604/781 [======================>.......] - ETA: 1s - loss: 1.5161 - accuracy: 0.4611


610/781 [======================>.......] - ETA: 1s - loss: 1.5162 - accuracy: 0.4612


616/781 [======================>.......] - ETA: 1s - loss: 1.5158 - accuracy: 0.4614


622/781 [======================>.......] - ETA: 1s - loss: 1.5151 - accuracy: 0.4615


628/781 [=======================>......] - ETA: 1s - loss: 1.5148 - accuracy: 0.4616


634/781 [=======================>......] - ETA: 1s - loss: 1.5139 - accuracy: 0.4619


640/781 [=======================>......] - ETA: 1s - loss: 1.5136 - accuracy: 0.4620


646/781 [=======================>......] - ETA: 1s - loss: 1.5135 - accuracy: 0.4622


652/781 [========================>.....] - ETA: 1s - loss: 1.5149 - accuracy: 0.4615


659/781 [========================>.....] - ETA: 1s - loss: 1.5145 - accuracy: 0.4615


666/781 [========================>.....] - ETA: 1s - loss: 1.5131 - accuracy: 0.4618


673/781 [========================>.....] - ETA: 0s - loss: 1.5121 - accuracy: 0.4621


680/781 [=========================>....] - ETA: 0s - loss: 1.5128 - accuracy: 0.4615


686/781 [=========================>....] - ETA: 0s - loss: 1.5125 - accuracy: 0.4617


693/781 [=========================>....] - ETA: 0s - loss: 1.5114 - accuracy: 0.4621


700/781 [=========================>....] - ETA: 0s - loss: 1.5114 - accuracy: 0.4622


707/781 [==========================>...] - ETA: 0s - loss: 1.5108 - accuracy: 0.4624


714/781 [==========================>...] - ETA: 0s - loss: 1.5108 - accuracy: 0.4626


721/781 [==========================>...] - ETA: 0s - loss: 1.5102 - accuracy: 0.4631


728/781 [==========================>...] - ETA: 0s - loss: 1.5103 - accuracy: 0.4631


735/781 [===========================>..] - ETA: 0s - loss: 1.5102 - accuracy: 0.4632


742/781 [===========================>..] - ETA: 0s - loss: 1.5103 - accuracy: 0.4631


749/781 [===========================>..] - ETA: 0s - loss: 1.5100 - accuracy: 0.4631


756/781 [============================>.] - ETA: 0s - loss: 1.5095 - accuracy: 0.4632


763/781 [============================>.] - ETA: 0s - loss: 1.5090 - accuracy: 0.4631


770/781 [============================>.] - ETA: 0s - loss: 1.5087 - accuracy: 0.4631


777/781 [============================>.] - ETA: 0s - loss: 1.5085 - accuracy: 0.4633


781/781 [==============================] - 7s 9ms/step - loss: 1.5088 - accuracy: 0.4632 - val_loss: 1.4748 - val_accuracy: 0.4753

Epoch 4/10

  1/781 [..............................] - ETA: 6s - loss: 1.6071 - accuracy: 0.3438


  8/781 [..............................] - ETA: 5s - loss: 1.4406 - accuracy: 0.5000


 15/781 [..............................] - ETA: 6s - loss: 1.4324 - accuracy: 0.4989


 22/781 [..............................] - ETA: 6s - loss: 1.4444 - accuracy: 0.4846


 29/781 [>.............................] - ETA: 6s - loss: 1.4586 - accuracy: 0.4923


 36/781 [>.............................] - ETA: 6s - loss: 1.4619 - accuracy: 0.4956


 42/781 [>.............................] - ETA: 6s - loss: 1.4750 - accuracy: 0.4898


 49/781 [>.............................] - ETA: 5s - loss: 1.4753 - accuracy: 0.4890


 56/781 [=>............................] - ETA: 5s - loss: 1.4868 - accuracy: 0.4850


 63/781 [=>............................] - ETA: 5s - loss: 1.4830 - accuracy: 0.4839


 70/781 [=>............................] - ETA: 5s - loss: 1.4786 - accuracy: 0.4831


 77/781 [=>............................] - ETA: 5s - loss: 1.4849 - accuracy: 0.4797


 84/781 [==>...........................] - ETA: 5s - loss: 1.4808 - accuracy: 0.4784


 91/781 [==>...........................] - ETA: 5s - loss: 1.4823 - accuracy: 0.4778


 98/781 [==>...........................] - ETA: 5s - loss: 1.4813 - accuracy: 0.4770


105/781 [===>..........................] - ETA: 5s - loss: 1.4788 - accuracy: 0.4780


112/781 [===>..........................] - ETA: 5s - loss: 1.4765 - accuracy: 0.4778


119/781 [===>..........................] - ETA: 5s - loss: 1.4741 - accuracy: 0.4794


126/781 [===>..........................] - ETA: 5s - loss: 1.4730 - accuracy: 0.4794


133/781 [====>.........................] - ETA: 5s - loss: 1.4693 - accuracy: 0.4810


140/781 [====>.........................] - ETA: 5s - loss: 1.4678 - accuracy: 0.4820


147/781 [====>.........................] - ETA: 5s - loss: 1.4704 - accuracy: 0.4810


153/781 [====>.........................] - ETA: 5s - loss: 1.4722 - accuracy: 0.4801


160/781 [=====>........................] - ETA: 4s - loss: 1.4685 - accuracy: 0.4816


167/781 [=====>........................] - ETA: 4s - loss: 1.4693 - accuracy: 0.4791


174/781 [=====>........................] - ETA: 4s - loss: 1.4716 - accuracy: 0.4790


181/781 [=====>........................] - ETA: 4s - loss: 1.4721 - accuracy: 0.4788


188/781 [======>.......................] - ETA: 4s - loss: 1.4737 - accuracy: 0.4779


195/781 [======>.......................] - ETA: 4s - loss: 1.4748 - accuracy: 0.4790


202/781 [======>.......................] - ETA: 4s - loss: 1.4736 - accuracy: 0.4802


209/781 [=======>......................] - ETA: 4s - loss: 1.4736 - accuracy: 0.4800


216/781 [=======>......................] - ETA: 4s - loss: 1.4748 - accuracy: 0.4798


223/781 [=======>......................] - ETA: 4s - loss: 1.4750 - accuracy: 0.4796


229/781 [=======>......................] - ETA: 4s - loss: 1.4756 - accuracy: 0.4788


236/781 [========>.....................] - ETA: 4s - loss: 1.4745 - accuracy: 0.4786


242/781 [========>.....................] - ETA: 4s - loss: 1.4742 - accuracy: 0.4782


249/781 [========>.....................] - ETA: 4s - loss: 1.4753 - accuracy: 0.4781


255/781 [========>.....................] - ETA: 4s - loss: 1.4757 - accuracy: 0.4779


262/781 [=========>....................] - ETA: 4s - loss: 1.4761 - accuracy: 0.4779


269/781 [=========>....................] - ETA: 4s - loss: 1.4746 - accuracy: 0.4784


276/781 [=========>....................] - ETA: 4s - loss: 1.4726 - accuracy: 0.4792


283/781 [=========>....................] - ETA: 4s - loss: 1.4723 - accuracy: 0.4796


290/781 [==========>...................] - ETA: 3s - loss: 1.4723 - accuracy: 0.4800


297/781 [==========>...................] - ETA: 3s - loss: 1.4725 - accuracy: 0.4799


304/781 [==========>...................] - ETA: 3s - loss: 1.4726 - accuracy: 0.4802


311/781 [==========>...................] - ETA: 3s - loss: 1.4717 - accuracy: 0.4803


318/781 [===========>..................] - ETA: 3s - loss: 1.4721 - accuracy: 0.4800


325/781 [===========>..................] - ETA: 3s - loss: 1.4731 - accuracy: 0.4796


332/781 [===========>..................] - ETA: 3s - loss: 1.4706 - accuracy: 0.4807


339/781 [============>.................] - ETA: 3s - loss: 1.4711 - accuracy: 0.4802


346/781 [============>.................] - ETA: 3s - loss: 1.4703 - accuracy: 0.4804


353/781 [============>.................] - ETA: 3s - loss: 1.4700 - accuracy: 0.4803


360/781 [============>.................] - ETA: 3s - loss: 1.4703 - accuracy: 0.4800


367/781 [=============>................] - ETA: 3s - loss: 1.4692 - accuracy: 0.4805


374/781 [=============>................] - ETA: 3s - loss: 1.4695 - accuracy: 0.4799


381/781 [=============>................] - ETA: 3s - loss: 1.4689 - accuracy: 0.4800


388/781 [=============>................] - ETA: 3s - loss: 1.4683 - accuracy: 0.4804


395/781 [==============>...............] - ETA: 3s - loss: 1.4685 - accuracy: 0.4805


402/781 [==============>...............] - ETA: 3s - loss: 1.4687 - accuracy: 0.4805


409/781 [==============>...............] - ETA: 2s - loss: 1.4685 - accuracy: 0.4804


416/781 [==============>...............] - ETA: 2s - loss: 1.4671 - accuracy: 0.4810


423/781 [===============>..............] - ETA: 2s - loss: 1.4674 - accuracy: 0.4806


429/781 [===============>..............] - ETA: 2s - loss: 1.4666 - accuracy: 0.4806


435/781 [===============>..............] - ETA: 2s - loss: 1.4671 - accuracy: 0.4802


442/781 [===============>..............] - ETA: 2s - loss: 1.4660 - accuracy: 0.4810


449/781 [================>.............] - ETA: 2s - loss: 1.4656 - accuracy: 0.4809


455/781 [================>.............] - ETA: 2s - loss: 1.4657 - accuracy: 0.4810


462/781 [================>.............] - ETA: 2s - loss: 1.4664 - accuracy: 0.4808


468/781 [================>.............] - ETA: 2s - loss: 1.4660 - accuracy: 0.4809


474/781 [=================>............] - ETA: 2s - loss: 1.4652 - accuracy: 0.4812


480/781 [=================>............] - ETA: 2s - loss: 1.4647 - accuracy: 0.4811


487/781 [=================>............] - ETA: 2s - loss: 1.4639 - accuracy: 0.4814


494/781 [=================>............] - ETA: 2s - loss: 1.4633 - accuracy: 0.4814


501/781 [==================>...........] - ETA: 2s - loss: 1.4625 - accuracy: 0.4820


508/781 [==================>...........] - ETA: 2s - loss: 1.4625 - accuracy: 0.4819


515/781 [==================>...........] - ETA: 2s - loss: 1.4616 - accuracy: 0.4821


522/781 [===================>..........] - ETA: 2s - loss: 1.4621 - accuracy: 0.4818


529/781 [===================>..........] - ETA: 2s - loss: 1.4626 - accuracy: 0.4818


536/781 [===================>..........] - ETA: 1s - loss: 1.4618 - accuracy: 0.4824


543/781 [===================>..........] - ETA: 1s - loss: 1.4620 - accuracy: 0.4825


550/781 [====================>.........] - ETA: 1s - loss: 1.4622 - accuracy: 0.4820


557/781 [====================>.........] - ETA: 1s - loss: 1.4617 - accuracy: 0.4819


564/781 [====================>.........] - ETA: 1s - loss: 1.4613 - accuracy: 0.4820


570/781 [====================>.........] - ETA: 1s - loss: 1.4621 - accuracy: 0.4818


577/781 [=====================>........] - ETA: 1s - loss: 1.4617 - accuracy: 0.4819


584/781 [=====================>........] - ETA: 1s - loss: 1.4621 - accuracy: 0.4819


591/781 [=====================>........] - ETA: 1s - loss: 1.4614 - accuracy: 0.4823


598/781 [=====================>........] - ETA: 1s - loss: 1.4610 - accuracy: 0.4822


604/781 [======================>.......] - ETA: 1s - loss: 1.4608 - accuracy: 0.4824


610/781 [======================>.......] - ETA: 1s - loss: 1.4611 - accuracy: 0.4826


616/781 [======================>.......] - ETA: 1s - loss: 1.4621 - accuracy: 0.4822


622/781 [======================>.......] - ETA: 1s - loss: 1.4615 - accuracy: 0.4825


628/781 [=======================>......] - ETA: 1s - loss: 1.4616 - accuracy: 0.4826


634/781 [=======================>......] - ETA: 1s - loss: 1.4608 - accuracy: 0.4829


641/781 [=======================>......] - ETA: 1s - loss: 1.4609 - accuracy: 0.4826


648/781 [=======================>......] - ETA: 1s - loss: 1.4608 - accuracy: 0.4826


655/781 [========================>.....] - ETA: 1s - loss: 1.4598 - accuracy: 0.4829


662/781 [========================>.....] - ETA: 0s - loss: 1.4602 - accuracy: 0.4827


669/781 [========================>.....] - ETA: 0s - loss: 1.4602 - accuracy: 0.4828


676/781 [========================>.....] - ETA: 0s - loss: 1.4601 - accuracy: 0.4826


683/781 [=========================>....] - ETA: 0s - loss: 1.4597 - accuracy: 0.4826


690/781 [=========================>....] - ETA: 0s - loss: 1.4590 - accuracy: 0.4830


696/781 [=========================>....] - ETA: 0s - loss: 1.4581 - accuracy: 0.4833


702/781 [=========================>....] - ETA: 0s - loss: 1.4578 - accuracy: 0.4831

---------------------------------------------------------------------------
KeyboardInterrupt                         Traceback (most recent call last)
/tmp/ipykernel_3563/2802866821.py in <module>
      2                   validation_data=(testX, testY),
      3                   steps_per_epoch=trainX.shape[0] // batch_size,
----> 4                   epochs=epochs, verbose=1)

/opt/hostedtoolcache/Python/3.7.13/x64/lib/python3.7/site-packages/keras/utils/traceback_utils.py in error_handler(*args, **kwargs)
     62     filtered_tb = None
     63     try:
---> 64       return fn(*args, **kwargs)
     65     except Exception as e:  # pylint: disable=broad-except
     66       filtered_tb = _process_traceback_frames(e.__traceback__)

/opt/hostedtoolcache/Python/3.7.13/x64/lib/python3.7/site-packages/keras/engine/training.py in fit(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, validation_batch_size, validation_freq, max_queue_size, workers, use_multiprocessing)
   1407                 _r=1):
   1408               callbacks.on_train_batch_begin(step)
-> 1409               tmp_logs = self.train_function(iterator)
   1410               if data_handler.should_sync:
   1411                 context.async_wait()

/opt/hostedtoolcache/Python/3.7.13/x64/lib/python3.7/site-packages/tensorflow/python/util/traceback_utils.py in error_handler(*args, **kwargs)
    148     filtered_tb = None
    149     try:
--> 150       return fn(*args, **kwargs)
    151     except Exception as e:
    152       filtered_tb = _process_traceback_frames(e.__traceback__)

/opt/hostedtoolcache/Python/3.7.13/x64/lib/python3.7/site-packages/tensorflow/python/eager/def_function.py in __call__(self, *args, **kwds)
    913 
    914       with OptionalXlaContext(self._jit_compile):
--> 915         result = self._call(*args, **kwds)
    916 
    917       new_tracing_count = self.experimental_get_tracing_count()

/opt/hostedtoolcache/Python/3.7.13/x64/lib/python3.7/site-packages/tensorflow/python/eager/def_function.py in _call(self, *args, **kwds)
    945       # In this case we have created variables on the first call, so we run the
    946       # defunned version which is guaranteed to never create variables.
--> 947       return self._stateless_fn(*args, **kwds)  # pylint: disable=not-callable
    948     elif self._stateful_fn is not None:
    949       # Release the lock early so that multiple threads can perform the call

/opt/hostedtoolcache/Python/3.7.13/x64/lib/python3.7/site-packages/tensorflow/python/eager/function.py in __call__(self, *args, **kwargs)
   2452        filtered_flat_args) = self._maybe_define_function(args, kwargs)
   2453     return graph_function._call_flat(
-> 2454         filtered_flat_args, captured_inputs=graph_function.captured_inputs)  # pylint: disable=protected-access
   2455 
   2456   @property

/opt/hostedtoolcache/Python/3.7.13/x64/lib/python3.7/site-packages/tensorflow/python/eager/function.py in _call_flat(self, args, captured_inputs, cancellation_manager)
   1859       # No tape is watching; skip to running the function.
   1860       return self._build_call_outputs(self._inference_function.call(
-> 1861           ctx, args, cancellation_manager=cancellation_manager))
   1862     forward_backward = self._select_forward_and_backward_functions(
   1863         args,

/opt/hostedtoolcache/Python/3.7.13/x64/lib/python3.7/site-packages/tensorflow/python/eager/function.py in call(self, ctx, args, cancellation_manager)
    500               inputs=args,
    501               attrs=attrs,
--> 502               ctx=ctx)
    503         else:
    504           outputs = execute.execute_with_cancellation(

/opt/hostedtoolcache/Python/3.7.13/x64/lib/python3.7/site-packages/tensorflow/python/eager/execute.py in quick_execute(op_name, num_outputs, inputs, attrs, ctx, name)
     53     ctx.ensure_initialized()
     54     tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,
---> 55                                         inputs, attrs, num_outputs)
     56   except core._NotOkStatusException as e:
     57     if name is not None:

KeyboardInterrupt: 

Guardando y recuperando el modelo

Necesitamos instalar

#!conda install -c anaconda pyyaml h5py

Guarda el modelo

Cuando corra el cuaderno localmente, use el siguiente comando para guardarlo

Python

mis_pesos = '../Checkpoints/Model_0_tf'
model.save_weights(mis_pesos)

Crea un nuevo modelo con mis pesos almacenados

Python

model_1 = NeuralNetwork()
model_1.compile(loss="categorical_crossentropy", optimizer=optimizer,
              metrics=["accuracy"])
model_1.load_weights(mis_pesos)

Evaluación del modelo

# Evalua el modelo original 
loss, acc = model.evaluate(testX, testY, verbose=2)

print("Exactitud del modelo original accuracy: {:5.2f}%".format(100 * acc))

313/313 - 3s - loss: 1.3708 - accuracy: 0.5143 - 3s/epoch - 9ms/step
Exactitud del modelo original accuracy: 51.43%

model_1=NeuralNetwork()
model_1.compile(loss="categorical_crossentropy", optimizer=optimizer,
              metrics=["accuracy"])

model_1.load_weights(model)

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
c:\Users\User\OneDrive\Documentos\Libro-Programacion\Tensorflow\Cuadernos\Tensorflow-01.ipynb Celda 32 in <cell line: 4>()
      <a href='vscode-notebook-cell:/c%3A/Users/User/OneDrive/Documentos/Libro-Programacion/Tensorflow/Cuadernos/Tensorflow-01.ipynb#ch0000040?line=0'>1</a> model_1=NeuralNetwork()
      <a href='vscode-notebook-cell:/c%3A/Users/User/OneDrive/Documentos/Libro-Programacion/Tensorflow/Cuadernos/Tensorflow-01.ipynb#ch0000040?line=1'>2</a> model_1.compile(loss="categorical_crossentropy", optimizer=optimizer,
      <a href='vscode-notebook-cell:/c%3A/Users/User/OneDrive/Documentos/Libro-Programacion/Tensorflow/Cuadernos/Tensorflow-01.ipynb#ch0000040?line=2'>3</a>               metrics=["accuracy"])
----> <a href='vscode-notebook-cell:/c%3A/Users/User/OneDrive/Documentos/Libro-Programacion/Tensorflow/Cuadernos/Tensorflow-01.ipynb#ch0000040?line=3'>4</a> model_1.load_weights(model)

File c:\Users\User\AppData\Local\Programs\Python\Python39\lib\site-packages\keras\utils\traceback_utils.py:67, in filter_traceback.<locals>.error_handler(*args, **kwargs)
     65 except Exception as e:  # pylint: disable=broad-except
     66   filtered_tb = _process_traceback_frames(e.__traceback__)
---> 67   raise e.with_traceback(filtered_tb) from None
     68 finally:
     69   del filtered_tb

File c:\Users\User\AppData\Local\Programs\Python\Python39\lib\site-packages\keras\saving\saving_utils.py:335, in is_hdf5_filepath(filepath)
    334 def is_hdf5_filepath(filepath):
--> 335   return (filepath.endswith('.h5') or filepath.endswith('.keras') or
    336           filepath.endswith('.hdf5'))

AttributeError: 'NeuralNetwork' object has no attribute 'endswith'

# Evalua el modelo restaurado
loss, acc = model_1.evaluate(testX, testY, verbose=2)

print("Exactitud del modelo restaurado accuracy: {:5.2f}%".format(100 * acc))

313/313 - 3s - loss: 1.3708 - accuracy: 0.5143 - 3s/epoch - 8ms/step
Exactitud del modelo restaurado accuracy: 51.43%

predictions = model.predict(testX)

313/313 [==============================] - 3s 10ms/step

Predicción

print(predictions[0])
print(np.argmax(predictions[0]))
print(testY[0])

[0.04558412 0.06122134 0.03857893 0.49717635 0.10958219 0.09021669
 0.01830608 0.00147913 0.13623646 0.00161872]
3
[0 0 0 1 0 0 0 0 0 0]

assert np.argmax(predictions[0]) == np.argmax(testY[0])