MCPcopy Create free account
hub / github.com/DeepRec-AI/DeepRec / main

Function main

modelzoo/bst/train.py:686–804  ·  view source on GitHub ↗
(tf_config=None, server=None)

Source from the content-addressed store, hash-verified

684
685
686def main(tf_config=None, server=None):
687 # check dataset and count data set size
688 print("Checking dataset...")
689 train_file = args.data_location + '/taobao_train_data'
690 test_file = args.data_location + '/taobao_test_data'
691 if args.parquet_dataset and not args.tf:
692 train_file += '.parquet'
693 test_file += '.parquet'
694 if (not os.path.exists(train_file)) or (not os.path.exists(test_file)):
695 print("Dataset does not exist in the given data_location.")
696 sys.exit()
697 no_of_training_examples = 0
698 no_of_test_examples = 0
699 if args.parquet_dataset and not args.tf:
700 import pyarrow.parquet as pq
701 no_of_training_examples = pq.read_table(train_file).num_rows
702 no_of_test_examples = pq.read_table(test_file).num_rows
703 else:
704 no_of_training_examples = sum(1 for line in open(train_file))
705 no_of_test_examples = sum(1 for line in open(test_file))
706 print("Numbers of training dataset is {}".format(no_of_training_examples))
707 print("Numbers of test dataset is {}".format(no_of_test_examples))
708
709 # set batch size, eporch & steps
710 batch_size = math.ceil(
711 args.batch_size / args.micro_batch
712 ) if args.micro_batch and not args.tf else args.batch_size
713
714 if args.steps == 0:
715 no_of_epochs = 100
716 train_steps = math.ceil(
717 (float(no_of_epochs) * no_of_training_examples) / batch_size)
718 else:
719 no_of_epochs = math.ceil(
720 (float(batch_size) * args.steps) / no_of_training_examples)
721 train_steps = args.steps
722 test_steps = math.ceil(float(no_of_test_examples) / batch_size)
723 print("The training steps is {}".format(train_steps))
724 print("The testing steps is {}".format(test_steps))
725
726 # set fixed random seed
727 tf.set_random_seed(args.seed)
728
729 # set directory path for checkpoint_dir
730 model_dir = os.path.join(args.output_dir,
731 'model_BST_' + str(int(time.time())))
732 checkpoint_dir = args.checkpoint if args.checkpoint else model_dir
733 print("Saving model checkpoints to " + checkpoint_dir)
734
735 # create data pipline of train & test dataset
736 train_dataset = build_model_input(train_file, batch_size, no_of_epochs)
737 test_dataset = build_model_input(test_file, batch_size, 1)
738
739 dataset_output_types = tf.data.get_output_types(train_dataset)
740 dataset_output_shapes = tf.data.get_output_shapes(test_dataset)
741 iterator = tf.data.Iterator.from_structure(dataset_output_types,
742 dataset_output_shapes)
743 next_element = iterator.get_next()

Callers 1

train.pyFile · 0.70

Calls 14

sumFunction · 0.85
BSTClass · 0.85
exitMethod · 0.80
timeMethod · 0.80
from_structureMethod · 0.80
build_model_inputFunction · 0.70
build_feature_columnsFunction · 0.70
trainFunction · 0.70
evalFunction · 0.70
formatMethod · 0.45
joinMethod · 0.45
get_nextMethod · 0.45

Tested by

no test coverage detected