def testMetric(self):
track.init(trial_name="test_log")
session = track.get_session()
for i in range(5):
track.log(test=i)
result_path = os.path.join(session.logdir, EXPR_RESULT_FILE)
self.assertTrue(_check_json_val(result_path, "test", i))
def train_mnist(args):
track.init(trial_name="track-example", trial_config=vars(args))
batch_size = 128
num_classes = 10
epochs = 1 if args.smoke_test else 12
mnist.load()
x_train, y_train, x_test, y_test, input_shape = get_mnist_data()
model = Sequential()
model.add(
Conv2D(32,
kernel_size=(3, 3),
activation="relu",
input_shape=input_shape))
model.add(Conv2D(64, (3, 3), activation="relu"))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.5))
model.add(Flatten())
model.add(Dense(args.hidden, activation="relu"))
model.add(Dropout(0.5))
model.add(Dense(num_classes, activation="softmax"))
model.compile(loss="categorical_crossentropy",
optimizer=keras.optimizers.SGD(lr=args.lr,
momentum=args.momentum),
metrics=["accuracy"])
model.fit(x_train,
y_train,
batch_size=batch_size,
epochs=epochs,
validation_data=(x_test, y_test),
callbacks=[TuneKerasCallback(track.metric)])
track.shutdown()
def train_mnist(args):
track.init(trial_name="track-example", trial_config=vars(args))
batch_size = 128
num_classes = 10
epochs = 1 if args.smoke_test else 12
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train, x_test = x_train / 255.0, x_test / 255.0
model = tf.keras.models.Sequential([
tf.keras.layers.Flatten(input_shape=(28, 28)),
tf.keras.layers.Dense(args.hidden, activation="relu"),
tf.keras.layers.Dropout(0.2),
tf.keras.layers.Dense(num_classes, activation="softmax")
])
model.compile(loss="sparse_categorical_crossentropy",
optimizer=keras.optimizers.SGD(lr=args.lr,
momentum=args.momentum),
metrics=["accuracy"])
model.fit(x_train,
y_train,
batch_size=batch_size,
epochs=epochs,
validation_data=(x_test, y_test),
callbacks=[TuneReporterCallback()])
track.shutdown()
def testLocalMetrics(self):
"""Checks that metric state is updated correctly."""
track.init(trial_name="test_logs")
session = track.get_session()
self.assertEqual(set(session.trial_config.keys()), {"trial_id"})
result_path = os.path.join(session.logdir, EXPR_RESULT_FILE)
track.log(test=1)
self.assertTrue(_check_json_val(result_path, "test", 1))
track.log(iteration=1, test=2)
self.assertTrue(_check_json_val(result_path, "test", 2))
def testSessionInitShutdown(self):
self.assertTrue(track._session is None)
# Checks that the singleton _session is created/destroyed
# by track.init() and track.shutdown()
for _ in range(2):
# do it twice to see that we can reopen the session
track.init(trial_name="test_init")
self.assertTrue(track._session is not None)
track.shutdown()
self.assertTrue(track._session is None)
def testLogCreation(self):
"""Checks that track.init() starts logger and creates log files."""
track.init(trial_name="test_init")
session = track.get_session()
self.assertTrue(session is not None)
self.assertTrue(os.path.isdir(session.logdir))
params_path = os.path.join(session.logdir, EXPR_PARAM_FILE)
result_path = os.path.join(session.logdir, EXPR_RESULT_FILE)
self.assertTrue(os.path.exists(params_path))
self.assertTrue(os.path.exists(result_path))
self.assertTrue(session.logdir == track.trial_dir())
def _trainable_func(self, config, reporter):
track.init(_tune_reporter=reporter)
output = train_func(config)
reporter(**{RESULT_DUPLICATE: True})
track.shutdown()
return output
analysis = tune.run(gridd,config={"lr": tune.grid_search([0.1,0.3])})
print("Best config: ",analysis.get_best_config(metric="mean_accuracy"))
analysis = tune.run(train,config={"num_epochs": tune.grid_search([5,10,15,20,25])})
print("Best config: ",analysis.get_best_config(metric="mean_accuracy"))
analysis = tune.run(train,config={"dropout": tune.grid_search([0.1,0.2,0.3])})
print("Best config: ",analysis.get_best_config(metric="mean_accuracy"))
analysis.dataframe()
# !ray
import ray
ray.tune.track.init()
ray.tune.track.init()
track.log
from ray import tune
import ray
from ray.tune import track
track.init()
