Detecta emocao notebook

In [ ]:

### USANDO O GOOGLE COLAB PARA BAIXAR O DATASET

import os

# Baixa o arquivo do Google Drive
!gdown 1MvvLoQzUbodQQ-eOCHU89LwX44gWxDgq -O utkface.zip

# Descompacta o zip
!unzip -q utkface.zip

# Descompacta o tar.gz
tar_path = '/content/fer2013.tar.gz'
extract_path = '/content/fer2013'
os.makedirs(extract_path, exist_ok=True)
!tar -xzf {tar_path} -C {extract_path}

# Caminho do CSV
csv_path = os.path.join(extract_path, 'fer2013/fer2013.csv')

### USANDO O GOOGLE COLAB PARA BAIXAR O DATASET import os # Baixa o arquivo do Google Drive !gdown 1MvvLoQzUbodQQ-eOCHU89LwX44gWxDgq -O utkface.zip # Descompacta o zip !unzip -q utkface.zip # Descompacta o tar.gz tar_path = '/content/fer2013.tar.gz' extract_path = '/content/fer2013' os.makedirs(extract_path, exist_ok=True) !tar -xzf {tar_path} -C {extract_path} # Caminho do CSV csv_path = os.path.join(extract_path, 'fer2013/fer2013.csv')

In [ ]:

import pandas as pd

data = pd.read_csv(csv_path)

data.head()

import pandas as pd data = pd.read_csv(csv_path) data.head()

Out[ ]:

	emotion	pixels	Usage
0	0	70 80 82 72 58 58 60 63 54 58 60 48 89 115 121...	Training
1	0	151 150 147 155 148 133 111 140 170 174 182 15...	Training
2	2	231 212 156 164 174 138 161 173 182 200 106 38...	Training
3	4	24 32 36 30 32 23 19 20 30 41 21 22 32 34 21 1...	Training
4	6	4 0 0 0 0 0 0 0 0 0 0 0 3 15 23 28 48 50 58 84...	Training

In [ ]:

from tensorflow.keras.utils import to_categorical
import numpy as np

# Função de pré-processamento
def preprocess_data(data):
    pixels = data['pixels'].tolist()
    faces = [np.fromstring(pixel_sequence, dtype=int, sep=' ').reshape(48, 48, 1) for pixel_sequence in pixels]
    faces = np.array(faces).astype('float32')
    faces /= 255.0
    emotions = to_categorical(data['emotion'], num_classes=7)
    return faces, emotions

# Separar os dados de treinamento, validação e teste
train_data = data[data['Usage'] == 'Training']
val_data = data[data['Usage'] == 'PublicTest']
test_data = data[data['Usage'] == 'PrivateTest']

x_train, y_train = preprocess_data(train_data)
x_val, y_val = preprocess_data(val_data)
x_test, y_test = preprocess_data(test_data)

# Verificar as formas dos dados
print('x_train shape:', x_train.shape)
print('y_train shape:', y_train.shape)
print('x_val shape:', x_val.shape)
print('y_val shape:', y_val.shape)
print('x_test shape:', x_test.shape)
print('y_test shape:', y_test.shape)

from tensorflow.keras.utils import to_categorical import numpy as np # Função de pré-processamento def preprocess_data(data): pixels = data['pixels'].tolist() faces = [np.fromstring(pixel_sequence, dtype=int, sep=' ').reshape(48, 48, 1) for pixel_sequence in pixels] faces = np.array(faces).astype('float32') faces /= 255.0 emotions = to_categorical(data['emotion'], num_classes=7) return faces, emotions # Separar os dados de treinamento, validação e teste train_data = data[data['Usage'] == 'Training'] val_data = data[data['Usage'] == 'PublicTest'] test_data = data[data['Usage'] == 'PrivateTest'] x_train, y_train = preprocess_data(train_data) x_val, y_val = preprocess_data(val_data) x_test, y_test = preprocess_data(test_data) # Verificar as formas dos dados print('x_train shape:', x_train.shape) print('y_train shape:', y_train.shape) print('x_val shape:', x_val.shape) print('y_val shape:', y_val.shape) print('x_test shape:', x_test.shape) print('y_test shape:', y_test.shape)

x_train shape: (28709, 48, 48, 1)
y_train shape: (28709, 7)
x_val shape: (3589, 48, 48, 1)
y_val shape: (3589, 7)
x_test shape: (3589, 48, 48, 1)
y_test shape: (3589, 7)

In [4]:

import matplotlib.pyplot as plt
import numpy as np

# Dicionário para mapear índices de emoção para rótulos textuais
label_to_text = {0: 'angry', 1: 'disgust', 2: 'fear', 3: 'happiness', 4: 'sad', 5: 'surprise', 6: 'neutral'}

# Função para exibir algumas imagens do dataset
def show_sample_images(faces, emotions, num_images=10):
    plt.figure(figsize=(15, 5))
    for i in range(num_images):
        ax = plt.subplot(2, num_images // 2, i + 1)
        plt.imshow(faces[i].reshape(48, 48), cmap='gray')
        plt.title(label_to_text[np.argmax(emotions[i])])
        plt.axis('off')
    plt.show()

# Exibir 10 imagens de exemplo do conjunto de treinamento
show_sample_images(x_train, y_train, num_images=10)

import matplotlib.pyplot as plt import numpy as np # Dicionário para mapear índices de emoção para rótulos textuais label_to_text = {0: 'angry', 1: 'disgust', 2: 'fear', 3: 'happiness', 4: 'sad', 5: 'surprise', 6: 'neutral'} # Função para exibir algumas imagens do dataset def show_sample_images(faces, emotions, num_images=10): plt.figure(figsize=(15, 5)) for i in range(num_images): ax = plt.subplot(2, num_images // 2, i + 1) plt.imshow(faces[i].reshape(48, 48), cmap='gray') plt.title(label_to_text[np.argmax(emotions[i])]) plt.axis('off') plt.show() # Exibir 10 imagens de exemplo do conjunto de treinamento show_sample_images(x_train, y_train, num_images=10)

In [7]:

from tensorflow.keras.preprocessing.image import ImageDataGenerator
# Criar geradores de dados
train_datagen = ImageDataGenerator(
    rotation_range=30,
    width_shift_range=0.2,
    height_shift_range=0.2,
    shear_range=0.2,
    zoom_range=0.2,
    horizontal_flip=True,
    fill_mode='nearest'
)

val_datagen = ImageDataGenerator()
test_datagen = ImageDataGenerator()

# Criar geradores
train_generator = train_datagen.flow(x_train, y_train, batch_size=64)
validation_generator = val_datagen.flow(x_val, y_val, batch_size=64)
test_generator = test_datagen.flow(x_test, y_test, batch_size=64)

from tensorflow.keras.preprocessing.image import ImageDataGenerator # Criar geradores de dados train_datagen = ImageDataGenerator( rotation_range=30, width_shift_range=0.2, height_shift_range=0.2, shear_range=0.2, zoom_range=0.2, horizontal_flip=True, fill_mode='nearest' ) val_datagen = ImageDataGenerator() test_datagen = ImageDataGenerator() # Criar geradores train_generator = train_datagen.flow(x_train, y_train, batch_size=64) validation_generator = val_datagen.flow(x_val, y_val, batch_size=64) test_generator = test_datagen.flow(x_test, y_test, batch_size=64)

In [8]:

from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Conv2D, MaxPooling2D, Flatten, Dense, Dropout

model = Sequential([

    Conv2D(32, kernel_size=(3, 3), activation='relu', input_shape=(48, 48, 1)),
    Conv2D(64, kernel_size=(3, 3), activation='relu'),
    MaxPooling2D(pool_size=(2, 2)),
    Dropout(0.2),

    Conv2D(128, kernel_size=(3, 3), activation='relu'),
    MaxPooling2D(pool_size=(2, 2)),
    Conv2D(128, kernel_size=(3, 3), activation='relu'),
    MaxPooling2D(pool_size=(2, 2)),
    Dropout(0.2),

    Flatten(),
    Dense(512, activation='relu'),
    Dropout(0.2),
    Dense(255, activation='relu'),
    Dropout(0.5),
    Dense(128, activation='relu'),
    Dropout(0.5),
    Dense(7, activation='softmax')
])

from tensorflow.keras.models import Sequential from tensorflow.keras.layers import Conv2D, MaxPooling2D, Flatten, Dense, Dropout model = Sequential([ Conv2D(32, kernel_size=(3, 3), activation='relu', input_shape=(48, 48, 1)), Conv2D(64, kernel_size=(3, 3), activation='relu'), MaxPooling2D(pool_size=(2, 2)), Dropout(0.2), Conv2D(128, kernel_size=(3, 3), activation='relu'), MaxPooling2D(pool_size=(2, 2)), Conv2D(128, kernel_size=(3, 3), activation='relu'), MaxPooling2D(pool_size=(2, 2)), Dropout(0.2), Flatten(), Dense(512, activation='relu'), Dropout(0.2), Dense(255, activation='relu'), Dropout(0.5), Dense(128, activation='relu'), Dropout(0.5), Dense(7, activation='softmax') ])

/usr/local/lib/python3.11/dist-packages/keras/src/layers/convolutional/base_conv.py:107: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.
  super().__init__(activity_regularizer=activity_regularizer, **kwargs)

In [9]:

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

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

In [10]:

from tensorflow.keras.callbacks import ModelCheckpoint, EarlyStopping, ReduceLROnPlateau


# Callbacks
checkpoint = ModelCheckpoint('best_model.h5', monitor='val_loss', save_best_only=True, mode='min')
early_stopping = EarlyStopping(monitor='val_loss', patience=10, restore_best_weights=True)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.001)

# Treinamento do modelo usando geradores
history = model.fit(
    train_generator,
    steps_per_epoch=len(x_train) // 64,
    epochs=500,
    validation_data=validation_generator,
    validation_steps=len(x_val) // 64,
    callbacks=[checkpoint, early_stopping, reduce_lr]
)

from tensorflow.keras.callbacks import ModelCheckpoint, EarlyStopping, ReduceLROnPlateau # Callbacks checkpoint = ModelCheckpoint('best_model.h5', monitor='val_loss', save_best_only=True, mode='min') early_stopping = EarlyStopping(monitor='val_loss', patience=10, restore_best_weights=True) reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.001) # Treinamento do modelo usando geradores history = model.fit( train_generator, steps_per_epoch=len(x_train) // 64, epochs=500, validation_data=validation_generator, validation_steps=len(x_val) // 64, callbacks=[checkpoint, early_stopping, reduce_lr] )

Epoch 1/500

/usr/local/lib/python3.11/dist-packages/keras/src/trainers/data_adapters/py_dataset_adapter.py:121: UserWarning: Your `PyDataset` class should call `super().__init__(**kwargs)` in its constructor. `**kwargs` can include `workers`, `use_multiprocessing`, `max_queue_size`. Do not pass these arguments to `fit()`, as they will be ignored.
  self._warn_if_super_not_called()

447/448 ━━━━━━━━━━━━━━━━━━━━ 0s 36ms/step - accuracy: 0.2316 - loss: 1.8410

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 31s 48ms/step - accuracy: 0.2317 - loss: 1.8409 - val_accuracy: 0.2489 - val_loss: 1.7968 - learning_rate: 0.0010
Epoch 2/500
  1/448 ━━━━━━━━━━━━━━━━━━━━ 3s 7ms/step - accuracy: 0.2969 - loss: 1.7708

/usr/local/lib/python3.11/dist-packages/keras/src/trainers/epoch_iterator.py:107: UserWarning: Your input ran out of data; interrupting training. Make sure that your dataset or generator can generate at least `steps_per_epoch * epochs` batches. You may need to use the `.repeat()` function when building your dataset.
  self._interrupted_warning()


448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 528us/step - accuracy: 0.2969 - loss: 1.7708 - val_accuracy: 0.2486 - val_loss: 1.7968 - learning_rate: 0.0010
Epoch 3/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.2533 - loss: 1.8024

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.2533 - loss: 1.8024 - val_accuracy: 0.2606 - val_loss: 1.7812 - learning_rate: 0.0010
Epoch 4/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 471us/step - accuracy: 0.2812 - loss: 1.7220 - val_accuracy: 0.2592 - val_loss: 1.7813 - learning_rate: 0.0010
Epoch 5/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.2519 - loss: 1.7956

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.2519 - loss: 1.7956 - val_accuracy: 0.2734 - val_loss: 1.7686 - learning_rate: 0.0010
Epoch 6/500
  1/448 ━━━━━━━━━━━━━━━━━━━━ 2s 5ms/step - accuracy: 0.2656 - loss: 1.7674

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 700us/step - accuracy: 0.2656 - loss: 1.7674 - val_accuracy: 0.2693 - val_loss: 1.7665 - learning_rate: 0.0010
Epoch 7/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 25ms/step - accuracy: 0.2581 - loss: 1.7811

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.2581 - loss: 1.7811 - val_accuracy: 0.3064 - val_loss: 1.7246 - learning_rate: 0.0010
Epoch 8/500
  1/448 ━━━━━━━━━━━━━━━━━━━━ 2s 5ms/step - accuracy: 0.2188 - loss: 1.7671

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 729us/step - accuracy: 0.2188 - loss: 1.7671 - val_accuracy: 0.3058 - val_loss: 1.7240 - learning_rate: 0.0010
Epoch 9/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.2694 - loss: 1.7648

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.2694 - loss: 1.7648 - val_accuracy: 0.3078 - val_loss: 1.6776 - learning_rate: 0.0010
Epoch 10/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 474us/step - accuracy: 0.2969 - loss: 1.7628 - val_accuracy: 0.3114 - val_loss: 1.6825 - learning_rate: 0.0010
Epoch 11/500
447/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.2817 - loss: 1.7338

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.2817 - loss: 1.7338 - val_accuracy: 0.3404 - val_loss: 1.6429 - learning_rate: 0.0010
Epoch 12/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 542us/step - accuracy: 0.2969 - loss: 1.7480 - val_accuracy: 0.3379 - val_loss: 1.6450 - learning_rate: 0.0010
Epoch 13/500
446/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.2946 - loss: 1.7198

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.2947 - loss: 1.7198 - val_accuracy: 0.3334 - val_loss: 1.6414 - learning_rate: 0.0010
Epoch 14/500
  1/448 ━━━━━━━━━━━━━━━━━━━━ 2s 5ms/step - accuracy: 0.3281 - loss: 1.7200

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 667us/step - accuracy: 0.3281 - loss: 1.7200 - val_accuracy: 0.3438 - val_loss: 1.6309 - learning_rate: 0.0010
Epoch 15/500
447/448 ━━━━━━━━━━━━━━━━━━━━ 0s 25ms/step - accuracy: 0.3049 - loss: 1.6943

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.3049 - loss: 1.6942 - val_accuracy: 0.3736 - val_loss: 1.5558 - learning_rate: 0.0010
Epoch 16/500
  1/448 ━━━━━━━━━━━━━━━━━━━━ 2s 5ms/step - accuracy: 0.2812 - loss: 1.7498

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 715us/step - accuracy: 0.2812 - loss: 1.7498 - val_accuracy: 0.3842 - val_loss: 1.5515 - learning_rate: 0.0010
Epoch 17/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.3359 - loss: 1.6607

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.3359 - loss: 1.6607 - val_accuracy: 0.4174 - val_loss: 1.4809 - learning_rate: 0.0010
Epoch 18/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 492us/step - accuracy: 0.3438 - loss: 1.6901 - val_accuracy: 0.4169 - val_loss: 1.4821 - learning_rate: 0.0010
Epoch 19/500
447/448 ━━━━━━━━━━━━━━━━━━━━ 0s 25ms/step - accuracy: 0.3635 - loss: 1.6056

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.3635 - loss: 1.6056 - val_accuracy: 0.4562 - val_loss: 1.4127 - learning_rate: 0.0010
Epoch 20/500
  1/448 ━━━━━━━━━━━━━━━━━━━━ 2s 5ms/step - accuracy: 0.4219 - loss: 1.4722

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 719us/step - accuracy: 0.4219 - loss: 1.4722 - val_accuracy: 0.4579 - val_loss: 1.4106 - learning_rate: 0.0010
Epoch 21/500
447/448 ━━━━━━━━━━━━━━━━━━━━ 0s 25ms/step - accuracy: 0.3798 - loss: 1.5749

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 11s 26ms/step - accuracy: 0.3798 - loss: 1.5749 - val_accuracy: 0.4754 - val_loss: 1.3532 - learning_rate: 0.0010
Epoch 22/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 501us/step - accuracy: 0.3906 - loss: 1.4922 - val_accuracy: 0.4704 - val_loss: 1.3552 - learning_rate: 0.0010
Epoch 23/500
447/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.4017 - loss: 1.5345

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4017 - loss: 1.5345 - val_accuracy: 0.4760 - val_loss: 1.3469 - learning_rate: 0.0010
Epoch 24/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 505us/step - accuracy: 0.4219 - loss: 1.5386 - val_accuracy: 0.4766 - val_loss: 1.3475 - learning_rate: 0.0010
Epoch 25/500
447/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.4209 - loss: 1.5013

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4208 - loss: 1.5013 - val_accuracy: 0.4830 - val_loss: 1.3233 - learning_rate: 0.0010
Epoch 26/500
  1/448 ━━━━━━━━━━━━━━━━━━━━ 2s 5ms/step - accuracy: 0.4688 - loss: 1.3883

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 719us/step - accuracy: 0.4688 - loss: 1.3883 - val_accuracy: 0.4858 - val_loss: 1.3172 - learning_rate: 0.0010
Epoch 27/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.4228 - loss: 1.4874

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.4228 - loss: 1.4874 - val_accuracy: 0.4967 - val_loss: 1.3062 - learning_rate: 0.0010
Epoch 28/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 488us/step - accuracy: 0.4531 - loss: 1.4086 - val_accuracy: 0.4908 - val_loss: 1.3116 - learning_rate: 0.0010
Epoch 29/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.4273 - loss: 1.4789

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.4273 - loss: 1.4789 - val_accuracy: 0.5075 - val_loss: 1.2880 - learning_rate: 0.0010
Epoch 30/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 514us/step - accuracy: 0.2969 - loss: 1.5920 - val_accuracy: 0.5050 - val_loss: 1.2881 - learning_rate: 0.0010
Epoch 31/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4367 - loss: 1.4683 - val_accuracy: 0.4947 - val_loss: 1.3114 - learning_rate: 0.0010
Epoch 32/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 499us/step - accuracy: 0.5000 - loss: 1.3740 - val_accuracy: 0.4972 - val_loss: 1.3071 - learning_rate: 0.0010
Epoch 33/500
447/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.4449 - loss: 1.4461

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4449 - loss: 1.4461 - val_accuracy: 0.5042 - val_loss: 1.2732 - learning_rate: 0.0010
Epoch 34/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 511us/step - accuracy: 0.5000 - loss: 1.4909 - val_accuracy: 0.5047 - val_loss: 1.2743 - learning_rate: 0.0010
Epoch 35/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.4414 - loss: 1.4490

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4414 - loss: 1.4490 - val_accuracy: 0.5162 - val_loss: 1.2642 - learning_rate: 0.0010
Epoch 36/500
  1/448 ━━━━━━━━━━━━━━━━━━━━ 2s 5ms/step - accuracy: 0.3750 - loss: 1.5260

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 724us/step - accuracy: 0.3750 - loss: 1.5260 - val_accuracy: 0.5181 - val_loss: 1.2631 - learning_rate: 0.0010
Epoch 37/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.4570 - loss: 1.4247

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.4570 - loss: 1.4247 - val_accuracy: 0.5134 - val_loss: 1.2589 - learning_rate: 0.0010
Epoch 38/500
  1/448 ━━━━━━━━━━━━━━━━━━━━ 2s 6ms/step - accuracy: 0.5312 - loss: 1.2810

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 716us/step - accuracy: 0.5312 - loss: 1.2810 - val_accuracy: 0.5145 - val_loss: 1.2578 - learning_rate: 0.0010
Epoch 39/500
447/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.4542 - loss: 1.4290

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4542 - loss: 1.4290 - val_accuracy: 0.5137 - val_loss: 1.2540 - learning_rate: 0.0010
Epoch 40/500
  1/448 ━━━━━━━━━━━━━━━━━━━━ 2s 6ms/step - accuracy: 0.5469 - loss: 1.2792

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 758us/step - accuracy: 0.5469 - loss: 1.2792 - val_accuracy: 0.5170 - val_loss: 1.2508 - learning_rate: 0.0010
Epoch 41/500
447/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.4572 - loss: 1.4035

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.4572 - loss: 1.4036 - val_accuracy: 0.5220 - val_loss: 1.2319 - learning_rate: 0.0010
Epoch 42/500
  1/448 ━━━━━━━━━━━━━━━━━━━━ 2s 6ms/step - accuracy: 0.3906 - loss: 1.5330

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 730us/step - accuracy: 0.3906 - loss: 1.5330 - val_accuracy: 0.5257 - val_loss: 1.2291 - learning_rate: 0.0010
Epoch 43/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4603 - loss: 1.4033 - val_accuracy: 0.5193 - val_loss: 1.2364 - learning_rate: 0.0010
Epoch 44/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 530us/step - accuracy: 0.3281 - loss: 1.4966 - val_accuracy: 0.5193 - val_loss: 1.2332 - learning_rate: 0.0010
Epoch 45/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.4624 - loss: 1.4050

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.4624 - loss: 1.4050 - val_accuracy: 0.5187 - val_loss: 1.2288 - learning_rate: 0.0010
Epoch 46/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 511us/step - accuracy: 0.5781 - loss: 1.2230 - val_accuracy: 0.5195 - val_loss: 1.2291 - learning_rate: 0.0010
Epoch 47/500
447/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.4648 - loss: 1.3957

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.4648 - loss: 1.3957 - val_accuracy: 0.5312 - val_loss: 1.2280 - learning_rate: 0.0010
Epoch 48/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 492us/step - accuracy: 0.4062 - loss: 1.5087 - val_accuracy: 0.5290 - val_loss: 1.2296 - learning_rate: 0.0010
Epoch 49/500
447/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.4677 - loss: 1.3843

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4677 - loss: 1.3843 - val_accuracy: 0.5360 - val_loss: 1.1992 - learning_rate: 0.0010
Epoch 50/500
  1/448 ━━━━━━━━━━━━━━━━━━━━ 2s 5ms/step - accuracy: 0.5156 - loss: 1.2925

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 729us/step - accuracy: 0.5156 - loss: 1.2925 - val_accuracy: 0.5360 - val_loss: 1.1984 - learning_rate: 0.0010
Epoch 51/500
447/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.4709 - loss: 1.3880

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.4709 - loss: 1.3880 - val_accuracy: 0.5449 - val_loss: 1.1979 - learning_rate: 0.0010
Epoch 52/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 501us/step - accuracy: 0.4844 - loss: 1.7526 - val_accuracy: 0.5463 - val_loss: 1.1981 - learning_rate: 0.0010
Epoch 53/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4784 - loss: 1.3806 - val_accuracy: 0.5382 - val_loss: 1.1983 - learning_rate: 0.0010
Epoch 54/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 478us/step - accuracy: 0.5312 - loss: 1.3241 - val_accuracy: 0.5377 - val_loss: 1.2000 - learning_rate: 0.0010
Epoch 55/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4721 - loss: 1.3732 - val_accuracy: 0.5315 - val_loss: 1.2094 - learning_rate: 0.0010
Epoch 56/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 498us/step - accuracy: 0.4375 - loss: 1.3556 - val_accuracy: 0.5301 - val_loss: 1.2099 - learning_rate: 0.0010
Epoch 57/500
447/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.4763 - loss: 1.3681

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4763 - loss: 1.3681 - val_accuracy: 0.5405 - val_loss: 1.1971 - learning_rate: 0.0010
Epoch 58/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 517us/step - accuracy: 0.4375 - loss: 1.5089 - val_accuracy: 0.5405 - val_loss: 1.2037 - learning_rate: 0.0010
Epoch 59/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4783 - loss: 1.3689 - val_accuracy: 0.5382 - val_loss: 1.2070 - learning_rate: 0.0010
Epoch 60/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 519us/step - accuracy: 0.3906 - loss: 1.5061 - val_accuracy: 0.5371 - val_loss: 1.2083 - learning_rate: 0.0010
Epoch 61/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.4899 - loss: 1.3485

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4899 - loss: 1.3485 - val_accuracy: 0.5444 - val_loss: 1.1810 - learning_rate: 0.0010
Epoch 62/500
  1/448 ━━━━━━━━━━━━━━━━━━━━ 2s 5ms/step - accuracy: 0.4844 - loss: 1.3667

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 715us/step - accuracy: 0.4844 - loss: 1.3667 - val_accuracy: 0.5463 - val_loss: 1.1809 - learning_rate: 0.0010
Epoch 63/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4854 - loss: 1.3486 - val_accuracy: 0.5366 - val_loss: 1.1987 - learning_rate: 0.0010
Epoch 64/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 500us/step - accuracy: 0.3906 - loss: 1.4372 - val_accuracy: 0.5366 - val_loss: 1.1986 - learning_rate: 0.0010
Epoch 65/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4824 - loss: 1.3554 - val_accuracy: 0.5416 - val_loss: 1.2121 - learning_rate: 0.0010
Epoch 66/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 507us/step - accuracy: 0.5938 - loss: 1.3561 - val_accuracy: 0.5416 - val_loss: 1.2089 - learning_rate: 0.0010
Epoch 67/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.4826 - loss: 1.3587

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.4826 - loss: 1.3586 - val_accuracy: 0.5480 - val_loss: 1.1722 - learning_rate: 0.0010
Epoch 68/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 513us/step - accuracy: 0.4375 - loss: 1.3768 - val_accuracy: 0.5446 - val_loss: 1.1758 - learning_rate: 0.0010
Epoch 69/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.4894 - loss: 1.3464 - val_accuracy: 0.5458 - val_loss: 1.1875 - learning_rate: 0.0010
Epoch 70/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 496us/step - accuracy: 0.5156 - loss: 1.4458 - val_accuracy: 0.5485 - val_loss: 1.1861 - learning_rate: 0.0010
Epoch 71/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4850 - loss: 1.3472 - val_accuracy: 0.5522 - val_loss: 1.1787 - learning_rate: 0.0010
Epoch 72/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 482us/step - accuracy: 0.5312 - loss: 1.2559 - val_accuracy: 0.5539 - val_loss: 1.1793 - learning_rate: 0.0010
Epoch 73/500
447/448 ━━━━━━━━━━━━━━━━━━━━ 0s 25ms/step - accuracy: 0.4888 - loss: 1.3482

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4888 - loss: 1.3481 - val_accuracy: 0.5527 - val_loss: 1.1647 - learning_rate: 0.0010
Epoch 74/500
  1/448 ━━━━━━━━━━━━━━━━━━━━ 2s 5ms/step - accuracy: 0.5312 - loss: 1.3610

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 683us/step - accuracy: 0.5312 - loss: 1.3610 - val_accuracy: 0.5544 - val_loss: 1.1634 - learning_rate: 0.0010
Epoch 75/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.4935 - loss: 1.3302

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4935 - loss: 1.3301 - val_accuracy: 0.5578 - val_loss: 1.1531 - learning_rate: 0.0010
Epoch 76/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 494us/step - accuracy: 0.5625 - loss: 1.2465 - val_accuracy: 0.5580 - val_loss: 1.1550 - learning_rate: 0.0010
Epoch 77/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.4902 - loss: 1.3386

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4902 - loss: 1.3386 - val_accuracy: 0.5647 - val_loss: 1.1510 - learning_rate: 0.0010
Epoch 78/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 530us/step - accuracy: 0.5312 - loss: 1.2438 - val_accuracy: 0.5647 - val_loss: 1.1551 - learning_rate: 0.0010
Epoch 79/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.4989 - loss: 1.3207 - val_accuracy: 0.5611 - val_loss: 1.1557 - learning_rate: 0.0010
Epoch 80/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 494us/step - accuracy: 0.5312 - loss: 1.3652 - val_accuracy: 0.5608 - val_loss: 1.1568 - learning_rate: 0.0010
Epoch 81/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.4939 - loss: 1.3305 - val_accuracy: 0.5572 - val_loss: 1.1630 - learning_rate: 0.0010
Epoch 82/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 556us/step - accuracy: 0.4531 - loss: 1.4783 - val_accuracy: 0.5552 - val_loss: 1.1636 - learning_rate: 0.0010
Epoch 83/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4932 - loss: 1.3353 - val_accuracy: 0.5572 - val_loss: 1.1605 - learning_rate: 0.0010
Epoch 84/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 539us/step - accuracy: 0.4844 - loss: 1.3984 - val_accuracy: 0.5580 - val_loss: 1.1586 - learning_rate: 0.0010
Epoch 85/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.5051 - loss: 1.3185

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5050 - loss: 1.3185 - val_accuracy: 0.5653 - val_loss: 1.1437 - learning_rate: 0.0010
Epoch 86/500
  1/448 ━━━━━━━━━━━━━━━━━━━━ 2s 5ms/step - accuracy: 0.5938 - loss: 1.2073

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 696us/step - accuracy: 0.5938 - loss: 1.2073 - val_accuracy: 0.5636 - val_loss: 1.1437 - learning_rate: 0.0010
Epoch 87/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.4969 - loss: 1.3222 - val_accuracy: 0.5664 - val_loss: 1.1506 - learning_rate: 0.0010
Epoch 88/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 501us/step - accuracy: 0.5000 - loss: 1.4272 - val_accuracy: 0.5695 - val_loss: 1.1510 - learning_rate: 0.0010
Epoch 89/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4974 - loss: 1.3223 - val_accuracy: 0.5597 - val_loss: 1.1438 - learning_rate: 0.0010
Epoch 90/500
  1/448 ━━━━━━━━━━━━━━━━━━━━ 2s 5ms/step - accuracy: 0.5938 - loss: 1.0789

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 672us/step - accuracy: 0.5938 - loss: 1.0789 - val_accuracy: 0.5569 - val_loss: 1.1417 - learning_rate: 0.0010
Epoch 91/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4996 - loss: 1.3272 - val_accuracy: 0.5575 - val_loss: 1.1460 - learning_rate: 0.0010
Epoch 92/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 507us/step - accuracy: 0.4062 - loss: 1.3718 - val_accuracy: 0.5594 - val_loss: 1.1474 - learning_rate: 0.0010
Epoch 93/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.4977 - loss: 1.3195 - val_accuracy: 0.5681 - val_loss: 1.1482 - learning_rate: 0.0010
Epoch 94/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 489us/step - accuracy: 0.4375 - loss: 1.4178 - val_accuracy: 0.5686 - val_loss: 1.1498 - learning_rate: 0.0010
Epoch 95/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.5098 - loss: 1.3130 - val_accuracy: 0.5639 - val_loss: 1.1609 - learning_rate: 0.0010
Epoch 96/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 483us/step - accuracy: 0.4844 - loss: 1.2244 - val_accuracy: 0.5619 - val_loss: 1.1648 - learning_rate: 0.0010
Epoch 97/500
447/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.5083 - loss: 1.3042

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.5083 - loss: 1.3042 - val_accuracy: 0.5670 - val_loss: 1.1373 - learning_rate: 0.0010
Epoch 98/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 542us/step - accuracy: 0.5676 - loss: 1.3522 - val_accuracy: 0.5647 - val_loss: 1.1381 - learning_rate: 0.0010
Epoch 99/500
447/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.5038 - loss: 1.3107

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5038 - loss: 1.3107 - val_accuracy: 0.5778 - val_loss: 1.1338 - learning_rate: 0.0010
Epoch 100/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 473us/step - accuracy: 0.4688 - loss: 1.4680 - val_accuracy: 0.5801 - val_loss: 1.1360 - learning_rate: 0.0010
Epoch 101/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5066 - loss: 1.3098 - val_accuracy: 0.5614 - val_loss: 1.1417 - learning_rate: 0.0010
Epoch 102/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 476us/step - accuracy: 0.4844 - loss: 1.2748 - val_accuracy: 0.5628 - val_loss: 1.1407 - learning_rate: 0.0010
Epoch 103/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5089 - loss: 1.3076 - val_accuracy: 0.5558 - val_loss: 1.1589 - learning_rate: 0.0010
Epoch 104/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 559us/step - accuracy: 0.4688 - loss: 1.3199 - val_accuracy: 0.5533 - val_loss: 1.1583 - learning_rate: 0.0010
Epoch 105/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5076 - loss: 1.3008 - val_accuracy: 0.5631 - val_loss: 1.1553 - learning_rate: 0.0010
Epoch 106/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 518us/step - accuracy: 0.5000 - loss: 1.5081 - val_accuracy: 0.5617 - val_loss: 1.1599 - learning_rate: 0.0010
Epoch 107/500
447/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.5096 - loss: 1.3022

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5096 - loss: 1.3022 - val_accuracy: 0.5723 - val_loss: 1.1293 - learning_rate: 0.0010
Epoch 108/500
  1/448 ━━━━━━━━━━━━━━━━━━━━ 2s 5ms/step - accuracy: 0.4531 - loss: 1.2126

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 690us/step - accuracy: 0.4531 - loss: 1.2126 - val_accuracy: 0.5700 - val_loss: 1.1291 - learning_rate: 0.0010
Epoch 109/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.5091 - loss: 1.2975

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.5091 - loss: 1.2975 - val_accuracy: 0.5642 - val_loss: 1.1267 - learning_rate: 0.0010
Epoch 110/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 469us/step - accuracy: 0.4219 - loss: 1.4107 - val_accuracy: 0.5614 - val_loss: 1.1277 - learning_rate: 0.0010
Epoch 111/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5069 - loss: 1.3058 - val_accuracy: 0.5678 - val_loss: 1.1361 - learning_rate: 0.0010
Epoch 112/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 476us/step - accuracy: 0.6250 - loss: 1.0984 - val_accuracy: 0.5672 - val_loss: 1.1361 - learning_rate: 0.0010
Epoch 113/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.5134 - loss: 1.2914

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5134 - loss: 1.2914 - val_accuracy: 0.5670 - val_loss: 1.1131 - learning_rate: 0.0010
Epoch 114/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 545us/step - accuracy: 0.3594 - loss: 1.4330 - val_accuracy: 0.5658 - val_loss: 1.1151 - learning_rate: 0.0010
Epoch 115/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.5129 - loss: 1.2867 - val_accuracy: 0.5742 - val_loss: 1.1205 - learning_rate: 0.0010
Epoch 116/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 516us/step - accuracy: 0.4375 - loss: 1.3622 - val_accuracy: 0.5717 - val_loss: 1.1190 - learning_rate: 0.0010
Epoch 117/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5125 - loss: 1.2864 - val_accuracy: 0.5711 - val_loss: 1.1141 - learning_rate: 0.0010
Epoch 118/500
  1/448 ━━━━━━━━━━━━━━━━━━━━ 2s 6ms/step - accuracy: 0.3594 - loss: 1.6869

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 698us/step - accuracy: 0.3594 - loss: 1.6869 - val_accuracy: 0.5714 - val_loss: 1.1121 - learning_rate: 0.0010
Epoch 119/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5133 - loss: 1.2935 - val_accuracy: 0.5706 - val_loss: 1.1265 - learning_rate: 0.0010
Epoch 120/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 535us/step - accuracy: 0.5312 - loss: 1.3436 - val_accuracy: 0.5720 - val_loss: 1.1277 - learning_rate: 0.0010
Epoch 121/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5169 - loss: 1.2861 - val_accuracy: 0.5714 - val_loss: 1.1316 - learning_rate: 0.0010
Epoch 122/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 519us/step - accuracy: 0.4844 - loss: 1.3687 - val_accuracy: 0.5695 - val_loss: 1.1366 - learning_rate: 0.0010
Epoch 123/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5206 - loss: 1.2850 - val_accuracy: 0.5583 - val_loss: 1.1430 - learning_rate: 0.0010
Epoch 124/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 478us/step - accuracy: 0.3906 - loss: 1.6088 - val_accuracy: 0.5575 - val_loss: 1.1425 - learning_rate: 0.0010
Epoch 125/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.5128 - loss: 1.2846

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5128 - loss: 1.2846 - val_accuracy: 0.5698 - val_loss: 1.1109 - learning_rate: 0.0010
Epoch 126/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 520us/step - accuracy: 0.5625 - loss: 1.2036 - val_accuracy: 0.5695 - val_loss: 1.1115 - learning_rate: 0.0010
Epoch 127/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5143 - loss: 1.2927 - val_accuracy: 0.5717 - val_loss: 1.1179 - learning_rate: 0.0010
Epoch 128/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 541us/step - accuracy: 0.5781 - loss: 1.2078 - val_accuracy: 0.5695 - val_loss: 1.1210 - learning_rate: 0.0010
Epoch 129/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.5182 - loss: 1.2823 - val_accuracy: 0.5723 - val_loss: 1.1224 - learning_rate: 0.0010
Epoch 130/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 479us/step - accuracy: 0.4531 - loss: 1.3808 - val_accuracy: 0.5720 - val_loss: 1.1216 - learning_rate: 0.0010
Epoch 131/500
447/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.5168 - loss: 1.2837

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.5169 - loss: 1.2837 - val_accuracy: 0.5795 - val_loss: 1.1088 - learning_rate: 0.0010
Epoch 132/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 519us/step - accuracy: 0.4062 - loss: 1.4958 - val_accuracy: 0.5767 - val_loss: 1.1112 - learning_rate: 0.0010
Epoch 133/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.5172 - loss: 1.2839 - val_accuracy: 0.5709 - val_loss: 1.1382 - learning_rate: 0.0010
Epoch 134/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 593us/step - accuracy: 0.5312 - loss: 1.3565 - val_accuracy: 0.5653 - val_loss: 1.1435 - learning_rate: 0.0010
Epoch 135/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5192 - loss: 1.2753 - val_accuracy: 0.5714 - val_loss: 1.1158 - learning_rate: 0.0010
Epoch 136/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 463us/step - accuracy: 0.5156 - loss: 1.2716 - val_accuracy: 0.5714 - val_loss: 1.1166 - learning_rate: 0.0010
Epoch 137/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 26ms/step - accuracy: 0.5239 - loss: 1.2672

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.5239 - loss: 1.2672 - val_accuracy: 0.5770 - val_loss: 1.1055 - learning_rate: 0.0010
Epoch 138/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 523us/step - accuracy: 0.5625 - loss: 1.2348 - val_accuracy: 0.5778 - val_loss: 1.1058 - learning_rate: 0.0010
Epoch 139/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5188 - loss: 1.2736 - val_accuracy: 0.5561 - val_loss: 1.1608 - learning_rate: 0.0010
Epoch 140/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 497us/step - accuracy: 0.5000 - loss: 1.1921 - val_accuracy: 0.5550 - val_loss: 1.1614 - learning_rate: 0.0010
Epoch 141/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5257 - loss: 1.2709 - val_accuracy: 0.5737 - val_loss: 1.1286 - learning_rate: 0.0010
Epoch 142/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 488us/step - accuracy: 0.5469 - loss: 1.2749 - val_accuracy: 0.5748 - val_loss: 1.1291 - learning_rate: 0.0010
Epoch 143/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5212 - loss: 1.2734 - val_accuracy: 0.5792 - val_loss: 1.1056 - learning_rate: 0.0010
Epoch 144/500
  1/448 ━━━━━━━━━━━━━━━━━━━━ 2s 5ms/step - accuracy: 0.5312 - loss: 1.2941

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 683us/step - accuracy: 0.5312 - loss: 1.2941 - val_accuracy: 0.5787 - val_loss: 1.1048 - learning_rate: 0.0010
Epoch 145/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5225 - loss: 1.2763 - val_accuracy: 0.5742 - val_loss: 1.1179 - learning_rate: 0.0010
Epoch 146/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 525us/step - accuracy: 0.5469 - loss: 1.2302 - val_accuracy: 0.5725 - val_loss: 1.1173 - learning_rate: 0.0010
Epoch 147/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5181 - loss: 1.2888 - val_accuracy: 0.5689 - val_loss: 1.1259 - learning_rate: 0.0010
Epoch 148/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 498us/step - accuracy: 0.5781 - loss: 1.2177 - val_accuracy: 0.5698 - val_loss: 1.1240 - learning_rate: 0.0010
Epoch 149/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5199 - loss: 1.2727 - val_accuracy: 0.5751 - val_loss: 1.1111 - learning_rate: 0.0010
Epoch 150/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 506us/step - accuracy: 0.5000 - loss: 1.4143 - val_accuracy: 0.5770 - val_loss: 1.1145 - learning_rate: 0.0010
Epoch 151/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5225 - loss: 1.2670 - val_accuracy: 0.5804 - val_loss: 1.1062 - learning_rate: 0.0010
Epoch 152/500
  1/448 ━━━━━━━━━━━━━━━━━━━━ 2s 6ms/step - accuracy: 0.5625 - loss: 1.2851

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. 


448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 732us/step - accuracy: 0.5625 - loss: 1.2851 - val_accuracy: 0.5809 - val_loss: 1.1037 - learning_rate: 0.0010
Epoch 153/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5284 - loss: 1.2679 - val_accuracy: 0.5776 - val_loss: 1.1213 - learning_rate: 0.0010
Epoch 154/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 514us/step - accuracy: 0.4688 - loss: 1.4563 - val_accuracy: 0.5739 - val_loss: 1.1200 - learning_rate: 0.0010
Epoch 155/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 27ms/step - accuracy: 0.5220 - loss: 1.2736 - val_accuracy: 0.5834 - val_loss: 1.1220 - learning_rate: 0.0010
Epoch 156/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 510us/step - accuracy: 0.4688 - loss: 1.4023 - val_accuracy: 0.5812 - val_loss: 1.1211 - learning_rate: 0.0010
Epoch 157/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5227 - loss: 1.2715 - val_accuracy: 0.5753 - val_loss: 1.1301 - learning_rate: 0.0010
Epoch 158/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 490us/step - accuracy: 0.5625 - loss: 1.1950 - val_accuracy: 0.5751 - val_loss: 1.1306 - learning_rate: 0.0010
Epoch 159/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5221 - loss: 1.2686 - val_accuracy: 0.5706 - val_loss: 1.1233 - learning_rate: 0.0010
Epoch 160/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 535us/step - accuracy: 0.4844 - loss: 1.3019 - val_accuracy: 0.5709 - val_loss: 1.1236 - learning_rate: 0.0010
Epoch 161/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 12s 26ms/step - accuracy: 0.5285 - loss: 1.2636 - val_accuracy: 0.5756 - val_loss: 1.1286 - learning_rate: 0.0010
Epoch 162/500
448/448 ━━━━━━━━━━━━━━━━━━━━ 0s 498us/step - accuracy: 0.5156 - loss: 1.2959 - val_accuracy: 0.5751 - val_loss: 1.1263 - learning_rate: 0.0010

In [11]:

# prompt: plote as curvas de loss e acuracia de treinameto do modelo

import matplotlib.pyplot as plt
# Plot the training and validation loss
plt.plot(history.history['loss'], label='Training Loss')
plt.plot(history.history['val_loss'], label='Validation Loss')
plt.title('Model Loss')
plt.ylabel('Loss')
plt.xlabel('Epoch')
plt.legend()
plt.show()

# Plot the training and validation accuracy
plt.plot(history.history['accuracy'], label='Training Accuracy')
plt.plot(history.history['val_accuracy'], label='Validation Accuracy')
plt.title('Model Accuracy')
plt.ylabel('Accuracy')
plt.xlabel('Epoch')
plt.legend()
plt.show()

# prompt: plote as curvas de loss e acuracia de treinameto do modelo import matplotlib.pyplot as plt # Plot the training and validation loss plt.plot(history.history['loss'], label='Training Loss') plt.plot(history.history['val_loss'], label='Validation Loss') plt.title('Model Loss') plt.ylabel('Loss') plt.xlabel('Epoch') plt.legend() plt.show() # Plot the training and validation accuracy plt.plot(history.history['accuracy'], label='Training Accuracy') plt.plot(history.history['val_accuracy'], label='Validation Accuracy') plt.title('Model Accuracy') plt.ylabel('Accuracy') plt.xlabel('Epoch') plt.legend() plt.show()

In [12]:

from sklearn.metrics import confusion_matrix, classification_report
import seaborn as sns


# Avaliação do modelo usando geradores
test_loss, test_acc = model.evaluate(x_test,y_test, steps=len(x_test) // 64)
print('Test accuracy:', test_acc)

# Previsões
y_pred = model.predict(x_test, steps=len(x_test) // 64)
y_pred_classes = np.argmax(y_pred, axis=1)
y_true = np.argmax(y_test, axis=1)

# Calcular a matriz de confusão
conf_matrix = confusion_matrix(y_true, y_pred_classes)

# Função para plotar a matriz de confusão
def plot_confusion_matrix(cm, classes, title='Confusion Matrix', cmap=plt.cm.Blues):
    plt.figure(figsize=(10, 8))
    sns.heatmap(cm, annot=True, fmt='d', cmap=cmap, xticklabels=classes, yticklabels=classes)
    plt.title(title)
    plt.ylabel('True Label')
    plt.xlabel('Predicted Label')
    plt.show()

# Plotar a matriz de confusão
plot_confusion_matrix(conf_matrix, classes=[label_to_text[i] for i in range(7)])

from sklearn.metrics import confusion_matrix, classification_report import seaborn as sns # Avaliação do modelo usando geradores test_loss, test_acc = model.evaluate(x_test,y_test, steps=len(x_test) // 64) print('Test accuracy:', test_acc) # Previsões y_pred = model.predict(x_test, steps=len(x_test) // 64) y_pred_classes = np.argmax(y_pred, axis=1) y_true = np.argmax(y_test, axis=1) # Calcular a matriz de confusão conf_matrix = confusion_matrix(y_true, y_pred_classes) # Função para plotar a matriz de confusão def plot_confusion_matrix(cm, classes, title='Confusion Matrix', cmap=plt.cm.Blues): plt.figure(figsize=(10, 8)) sns.heatmap(cm, annot=True, fmt='d', cmap=cmap, xticklabels=classes, yticklabels=classes) plt.title(title) plt.ylabel('True Label') plt.xlabel('Predicted Label') plt.show() # Plotar a matriz de confusão plot_confusion_matrix(conf_matrix, classes=[label_to_text[i] for i in range(7)])

56/56 ━━━━━━━━━━━━━━━━━━━━ 10s 66ms/step - accuracy: 0.5816 - loss: 1.0889
Test accuracy: 0.591251015663147
56/56 ━━━━━━━━━━━━━━━━━━━━ 2s 22ms/step

In [13]:

# Exibir 10 imagens de exemplo do conjunto de teste
show_sample_images(x_test, y_pred, num_images=10)

# Exibir 10 imagens de exemplo do conjunto de teste show_sample_images(x_test, y_pred, num_images=10)