import json
import os
import datetime
from model_module import ModelBuilder
from ResNetTester_cls import ResNetTester, make, check_run
from Datasets import (cifer10_datasets, cifer100_datasets, mnist_dataset)
from tools import converter_json
global_name = "20190930_test"
loop = 3
relu_option = False
epochs = 1
split = 1.0
batch_size = 32
dataset = cifer10_datasets(is_zero_center=True)
json_file = "options/20190930_options_bk.json"
result_file = []
for i in range(loop):
global_loop_name = global_name + str(i)
json_experience = ""
with open(json_file) as f:
s = f.read()
json_experience = json.loads(s)
for e in json_experience["options"]:
filename = 'result/' + global_loop_name + '.json'
isRuned = check_run(e, filename)
result_file.append(filename)
'''
それぞれのネットワークの評価コード
'''
import numpy as np
import sklearn.metrics as metrics
from ModelBuilder import ResnetBuilder
from plot_result import (plot_acc_history, plot_loss_history,
outputfile_evalute, get_time)
from ResNetTester_cls import ResNetTester
from Datasets import cifer10_datasets
data = cifer10_datasets()
Testers = [
ResNetTester('ResNet18'),
ResNetTester('Invert_ResNet18'),
ResNetTester('Dual_ReLU_ResNet18'),
ResNetTester('Dual_ReLU_Concatenate_ResNet18')
]
output_log_file = get_time() + '_' + 'result.txt'
for index, test in enumerate(Testers):
test.setDataset(data)
if index == 0:
model = ResnetBuilder.build_resnet_18(data.init_shape, 10)
elif index == 1:
model = ResnetBuilder.build_invert_relu_resnet_18(data.init_shape, 10)
elif index == 2:
model = ResnetBuilder.build_dualresnet_18(data.init_shape, 10)
elif index == 3:
#from ModelBuilder import ResnetBuilder
from model_module import ModelBuilder
from Datasets import (cifer10_datasets, cifer100_datasets, mnist_dataset)
from keras.utils import plot_model
dataset = cifer10_datasets(is_zero_center=False)
option = {
"relu_option": False,
"double_input": False,
"concatenate": "none",
"block": "basic_block",
"reseption": [2, 2, 2, 2],
"dropout": 0,
"wide": False,
"filters": 64
}
model = ModelBuilder.ResnetBuilder.build_manual(dataset.get_shape(),
dataset.get_categorical(),
option)
model.summary()
plot_model(model, to_file='model.png', show_shapes=True, show_layer_names=True)