sys.exit(0)
if args.submit == 'nni':
builder.generate_code()
tunor = builder.find_role('hyper-tunor', unique=True)
commands = [
'nnictl', 'create', '--config',
tunor.store_nni_config()
]
builder.node.submit_job(commands)
sys.exit(0)
if args.nni_trial:
import nni
builder.stages[0].update_trial_id('trial-' + random_string(size=8))
param = nni.get_parameters()
print(param)
builder.update_config(param)
result = builder.build_and_train()
# nni.report_final_result(0 if result is None else result['loss'])
sys.exit(0)
# build
if args.train:
builder = ModelPipeline(config, from_fe=args.front_end)
builder.build_and_train()
builder.notify('Succeed', 'Trainig has been done')
sys.exit(0)
except Exception as e:
builder.notify(
'Failed',
model.weights = PARAMS['knr_weights']
except Exception as exception:
LOG.exception(exception)
raise
return model
def run(X_train, X_test, y_train, y_test, PARAMS):
'''Train model and predict result'''
model.fit(X_train, y_train)
predict_y = model.predict(X_test)
score = r2_score(y_test, predict_y)
LOG.debug('r2 score: %s' % score)
nni.report_final_result(score)
if __name__ == '__main__':
X_train, X_test, y_train, y_test = load_data()
try:
# get parameters from tuner
RECEIVED_PARAMS = nni.get_parameters()
LOG.debug(RECEIVED_PARAMS)
PARAMS = get_default_parameters()
PARAMS.update(RECEIVED_PARAMS)
LOG.debug(PARAMS)
model = get_model(PARAMS)
run(X_train, X_test, y_train, y_test, model)
except Exception as exception:
LOG.exception(exception)
raise
save_path = None
max_epoch = args.max_epoch
cfg = GAGConfig()
cfg.batch_size = args.batch_size
cfg.learning_rate = float(args.learning_rate)
cfg.dropout = args.dropout_rate
cfg.rnn_units = args.rnn_units
cfg.labelsmoothing = args.labelsmoothing
cfg.num_heads = args.num_heads
timer = Timer()
qp_pairs, dev_qp_pairs = load_data()
logger.debug('Init finish.')
original_params = nni.get_parameters()
'''
with open('data.json') as f:
original_params = json.load(f)
'''
try:
graph = graph.graph_loads(original_params)
except Exception:
logger.debug('Can\'t load graph.')
train_loss, best_acc = train_with_graph(graph, qp_pairs, dev_qp_pairs)
logger.debug('Send best acc: %s', str(best_acc))
nni.report_final_result(best_acc)
logger.debug('Send final result done')
except:
logger.exception('Catch exception in trial.py.')
def generate_default_params():
'''
Generate default parameters for mnist network.
'''
params = {
'data_dir': '/tmp/tensorflow/mnist/input_data',
'dropout_rate': 0.5,
'channel_1_num': 32,
'channel_2_num': 64,
'conv_size': 5,
'pool_size': 2,
'hidden_size': 1024,
'learning_rate': 1e-4,
'batch_num': 200}
return params
if __name__ == '__main__':
try:
# get parameters form tuner
RCV_PARAMS = nni.get_parameters()
logger.debug(RCV_PARAMS)
# run
params = generate_default_params()
params.update(RCV_PARAMS)
main(params)
except Exception as exception:
logger.exception(exception)
raise
import nni
params = nni.get_parameters()
print('params:', params)
x = params['x']
nni.report_final_result(x)
print('Saving..')
state = {
'net': net.state_dict(),
'acc': acc,
'epoch': epoch,
}
if not os.path.isdir('checkpoint'):
os.mkdir('checkpoint')
torch.save(state, './checkpoint/ckpt.t7')
best_acc = acc
return acc, best_acc
if __name__ == '__main__':
try:
RCV_CONFIG = nni.get_parameters()
#RCV_CONFIG = {'lr': 0.1, 'optimizer': 'Adam', 'model':'senet18'}
_logger.debug(RCV_CONFIG)
prepare(RCV_CONFIG)
acc = 0.0
best_acc = 0.0
for epoch in range(start_epoch, start_epoch+200):
train(epoch)
acc, best_acc = test(epoch)
nni.report_intermediate_result(acc)
nni.report_final_result(best_acc)
except Exception as exception:
_logger.exception(exception)
raise
return params
def parse_init_json(data):
params = {}
for key in data:
value = data[key]
if value == 'Empty':
params[key] = ['Empty']
else:
params[key] = [value[0], value[1]['_value'], value[1]['_value']]
return params
if __name__ == '__main__':
try:
# get parameters form tuner
data = nni.get_parameters()
logger.debug(data)
RCV_PARAMS = parse_init_json(data)
logger.debug(RCV_PARAMS)
params = generate_defualt_params()
params.update(RCV_PARAMS)
print(RCV_PARAMS)
main(params)
except Exception as exception:
logger.exception(exception)
raise