def _report_parameter_from_model_optimizer(self, metric_name, name_of_optimizer_attr=None):
if not name_of_optimizer_attr:
name_of_optimizer_attr = metric_name
if not _should_report_metric_or_parameter(autolog_inputs, metric_name) or not hasattr(self.model.optimizer, name_of_optimizer_attr):
return
parameter_from_model = getattr(self.model.optimizer, name_of_optimizer_attr)
value = parameter_from_model if type(parameter_from_model) is float else keras.backend.eval(parameter_from_model)
reportParameter(metric_name, value)
import keras
from keras.datasets import mnist
from keras.models import Sequential
from keras.layers import Dense, Dropout
import time
from runai.reporter import autolog, reportParameter, reportMetric
import keras.optimizers
NUM_CLASSES = 10
BATCH_SIZE = 10
STEPS = 5
autolog()
reportParameter("current_state", "preprocessing")
# the data, split between train and test sets
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train = x_train.reshape(60000, 784)
x_test = x_test.reshape(10000, 784)
x_train = x_train.astype('float32')
x_test = x_test.astype('float32')
x_train /= 255
x_test /= 255
reportParameter("current_state", "final tunning")
# convert class vectors to binary class matrices
y_train = keras.utils.to_categorical(y_train, NUM_CLASSES)
y_test = keras.utils.to_categorical(y_test, NUM_CLASSES)
def _reportParameter(*args, **kwargs):
if reporter is not None:
reporter.reportParameter(*args, **kwargs)
else:
runai.reporter.reportParameter(*args, **kwargs)
from runai.reporter import reportMetric, reportParameter
from time import sleep
for step in range(1000):
reportMetric("step", step)
reportMetric("accuracy", step)
reportMetric("loss", step)
reportMetric("epoch", step)
reportMetric("this_is_a_test", 123)
reportParameter("state", "running")
sleep(1)
def _report_parameter_if_needed(autolog_inputs, key, value):
if _should_report_metric_or_parameter(autolog_inputs, key):
reportParameter(key, value)