def train(params):
x, labels, sess, train_op, accuracy, loss, merged, train_writer, gst = model_fn(params, mode='train')
xdata, labeldata = train_input_fn()
print('Training begins')
start_time=time.time()
for j in range (params.num_epochs):
print("EPOCH NUMBER: ", j+1)
avg_acc=0
avg_lss=0
for k in range(0, len(xdata), params.batch_size):
current_batch_x_train = xdata[k:k+params.batch_size]
current_batch_label_train = labeldata[k:k+params.batch_size]
_, acc, lss, merg, gg= sess.run([train_op, accuracy, loss, merged, gst],
feed_dict = {x: current_batch_x_train, labels: current_batch_label_train})
avg_acc+=acc
avg_lss+=lss
#print("Mini-batch training Accuracy", acc)
#print("Mini-batch training Loss", lss)
train_writer.add_summary(merg, global_step=gg)
print("Average Training Accuracy= ", avg_acc * params.batch_size/len(xdata))
print("Average Training Loss= ", avg_lss * params.batch_size/len(xdata))
print("xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")
train_time=time.time() - start_time
saver= tf.train.Saver()
saver.save(sess , os.path.join(os.getcwd(), params.model_file))
print("Total training time= ", train_time, "seconds")
ANCHORS = [0, 1, 2] # 0 for "tweet", 1 for "image", 2 for "user"
if __name__ == '__main__':
tf.reset_default_graph()
#tf.logging.set_verbosity(tf.logging.INFO)
# Load the parameters from json file
args = parser.parse_args()
json_path = os.path.join(args.model_dir, 'params.json')
assert os.path.isfile(json_path), "No json configuration file found at {}".format(json_path)
params = Params(json_path)
# Get the datasets
#tf.logging.info("Getting the dataset...")
dataset_iter = train_input_fn(args.data_dir, params, args.embed_dir)
dataset_next = dataset_iter.get_next()
# Define the model
#tf.logging.info("Creating the model...")
model = TripletLoss(params)
num_train = 5031
num_train_steps = int(num_train/params.batch_size) * params.num_epochs
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
sess.run(tf.local_variables_initializer())
""" Training Module """
for i in tqdm(range(0, num_train_steps)):
sess.run(dataset_iter.initializer)
# Load the parameters from json file
args = parser.parse_args()
json_path = os.path.join(args.model_dir, 'params.json')
assert os.path.isfile(json_path), "No json configuration file found at {}".format(json_path)
params = Params(json_path)
# Define the model
tf.logging.info("Creating the model...")
config = tf.estimator.RunConfig(tf_random_seed=230,
model_dir=args.model_dir,
save_summary_steps=params.save_summary_steps)
estimator = tf.estimator.Estimator(model_fn, params=params, config=config)
# Train the model
tf.logging.info("Starting training for {} epoch(s).".format(params.num_epochs))
estimator.train(lambda: train_input_fn(args.data_dir, params))
'''
path = args.model_save
try:
os.makedirs(path)
except OSError:
print ("Creation of the directory %s failed" % path)
else:
print ("Successfully created the directory %s " % path)
estimator.export_saved_model(path, serving_input_receiver_fn())
print("Model is SAVED !!")
args = parser.parse_args()
json_path = os.path.join(args.model_dir, 'params.json')
assert os.path.isfile(
json_path), "No json configuration file found at {}".format(json_path)
params = Params(json_path)
# Define the model
tf.logging.info("Creating the model...")
config = tf.estimator.RunConfig(
tf_random_seed=230,
model_dir=args.model_dir,
save_summary_steps=params.save_summary_steps)
estimator = tf.estimator.Estimator(model_fn, params=params, config=config)
# Train the model
tf.logging.info("Starting training for {} epoch(s).".format(
params.num_epochs))
start = time.clock()
estimator.train(
lambda: train_input_fn(args.data_dir, params, args.model_dir))
tf.logging.info("Training took %s seconds." % (time.clock() - start))
# Evaluate the model on the validation set
tf.logging.info("Evaluation on validation set.")
start = time.clock()
res = estimator.evaluate(
lambda: val_input_fn(args.data_dir, params, args.model_dir))
tf.logging.info("Evaluation took %s seconds." % (time.clock() - start))
for key in res:
print("{}: {}".format(key, res[key]))
import tensorflow as tf
from model.model import create_model
from model.input_fn import train_input_fn
from model.loss import composite_loss
from model.utils import Vocabulary
import os, json
ROOT_PATH = os.path.dirname(os.path.abspath(__file__))
STATIC_CONFIG = dict(json.load(open('config.json', 'r')))
print(staticmethod)
RUNTIME_CONFIG = {"root_path": ROOT_PATH}
print(RUNTIME_CONFIG)
CONFIG = {**STATIC_CONFIG, **RUNTIME_CONFIG}
vocabulary = Vocabulary()
next_training_batch = train_input_fn(vocabulary, CONFIG)
modeloutput - create_model(next_training_batch, CONFIG, True)
