(comm_block.Get_size() - 1),
elastic_lr=args.elastic_lr,
elastic_momentum=args.elastic_momentum)
else:
algo = Algo(args.optimizer,
loss=args.loss,
validate_every=validate_every,
sync_every=args.sync_every,
worker_optimizer=args.worker_optimizer)
os.environ['KERAS_BACKEND'] = backend
#import_keras()
block = process_block.ProcessBlock(comm_world,
comm_block,
algo,
data,
device,
model_provider,
args.epochs,
train_list,
val_list,
folds=args.n_fold,
num_masters=args.n_master,
num_process=args.n_process,
verbose=args.verbose,
early_stopping=args.early_stopping,
target_metric=args.target_metric,
monitor=args.monitor)
block.label = args.label
block.run()
param_ranges, model_fn)
opt_coordinator.run(num_iterations=30)
else:
data = H5Data(batch_size=args.batch,
features_name='Images',
labels_name='Labels')
data.set_file_names(train_list)
validate_every = data.count_data() / args.batch
algo = Algo(args.optimizer,
loss=args.loss,
validate_every=validate_every,
sync_every=args.sync_every)
os.environ['KERAS_BACKEND'] = backend
import_keras()
import keras.callbacks as cbks
callbacks = []
if args.early_stopping is not None:
callbacks.append(
cbks.EarlyStopping(patience=args.early_stopping, verbose=1))
block = process_block.ProcessBlock(comm_world,
comm_block,
algo,
data,
device,
args.epochs,
train_list,
val_list,
callbacks,
verbose=args.verbose)
block.run()