print("-----=-=-=- checkpoint_loaded=-=-=--------------")
try:
history = model.fit(X_train,
y_train,
batch_size=batch_size,
epochs=epochs,
callbacks=[checkpoint, callbacks],
validation_data=(X_valid, y_valid),
verbose=1)
except Exception as e:
runtimeInvalidInfo = {
'calculationType': 'exception',
'info': str(format(e))
}
response = http_client.modelTrain(str(runtimeInvalidInfo))
print("un expected error").format(e)
# 指标返回
call_res = call_back_metrics(X_train, X_valid, X_test, y_train, y_valid,
y_test, model, local_img_path)
# 回调，向服务端发送评估指标
try:
response = http_client.modelTrain(str(call_res))
except Exception as e:
print("un expected error").format(e)
# http_client.call("sayHelloWorld",call_res)
model.save(os.path.join(save_dir, model_name) + '.h5', overwrite=True)
# Fit the model
if os.path.exists(filepath):
print('now loadding the weights file ', filepath)
model.load_weights(filepath)
# 若成功加载前面保存的参数，输出下列信息
print("-----=-=-=- checkpoint_loaded=-=-=--------------")
history = model.fit(
X_train,
y_train,
batch_size=batch_size,
epochs=epochs,
# callbacks=callbacks,
callbacks=[cb],
validation_data=(X_valid, y_valid),
verbose=1)
# 指标返回
call_res = call_back_metrics(X_train, X_valid, X_test, y_train, y_valid,
y_test, model)
# 回调，向服务端发送评估指标
response = http_client.modelTrain(str(call_res))
# http_client.call("sayHelloWorld",call_res)
model.save(os.path.join(save_dir, model_name) + '.h5', overwrite=True)
print('\r\nmodel has been saved in ',
os.path.join(save_dir, model_name) + '.h5')
