' DMN type is not currently implemented')
if args.babi_task_id is not None:
config.babi_id = args.babi_task_id
config.strong_supervision = False
config.train_mode = False
print('Testing DMN ' + dmn_type + ' on babi task', config.babi_id)
# create model
with tf.variable_scope('DMN') as scope:
if dmn_type == "original":
from dmn_original import DMN
model = DMN(config)
elif dmn_type == "plus":
from dmn_plus import DMN_PLUS
model = DMN_PLUS(config)
print('==> initializing variables')
init = tf.global_variables_initializer()
saver = tf.train.Saver()
with tf.Session() as session:
session.run(init)
print('==> restoring weights')
saver.restore(session,
'weights/task' + str(model.config.babi_id) + '.weights')
print('==> running DMN')
config.babi_id = args.babi_task_id
config.babi_id = args.babi_task_id if args.babi_task_id is not None else str(1)
config.l2 = args.l2_loss if args.l2_loss is not None else 0.001
config.strong_supervision = args.strong_supervision if args.strong_supervision is not None else False
num_runs = args.num_runs if args.num_runs is not None else 1
print 'Training DMN ' + dmn_type + ' on babi task', config.babi_id
best_overall_val_loss = float('inf')
# create model
with tf.variable_scope('DMN') as scope:
if dmn_type == "original":
from dmn_original import DMN
model = DMN(config)
elif dmn_type == "plus":
from dmn_plus import DMN_PLUS
model = DMN_PLUS(config)
for run in range(num_runs):
print 'Starting run', run
print '==> initializing variables'
init = tf.initialize_all_variables()
saver = tf.train.Saver()
with tf.Session() as session:
sum_dir = 'summaries/train/' + time.strftime("%Y-%m-%d %H %M")
raise NotImplementedError(dmn_type + ' DMN type is not currently implemented')
if args.babi_task_id is not None:
config.babi_id = args.babi_task_id
config.strong_supervision = False
config.train_mode = False
print( 'Testing DMN ' + dmn_type + ' on babi task', config.babi_id)
# create model
with tf.variable_scope('DMN') as scope:
if dmn_type == "original":
from dmn_original import DMN
model = DMN(config)
elif dmn_type == "plus":
from dmn_plus import DMN_PLUS
model = DMN_PLUS(config)
print('==> initializing variables')
init = tf.global_variables_initializer()
saver = tf.train.Saver()
with tf.Session() as session:
session.run(init)
print('==> restoring weights')
saver.restore(session, 'weights/task' + str(model.config.babi_id) + '.weights')
print('==> running DMN')
config.babi_id = args.babi_task_id
config.babi_id = args.babi_task_id if args.babi_task_id is not None else str(1)
config.l2 = args.l2_loss if args.l2_loss is not None else 0.001
config.strong_supervision = args.strong_supervision if args.strong_supervision is not None else False
num_runs = args.num_runs if args.num_runs is not None else 1
print('Training DMN ' + dmn_type + ' on babi task', config.babi_id)
best_overall_val_loss = float('inf')
# create model
with tf.variable_scope('DMN') as scope:
if dmn_type == "original":
from dmn_original import DMN
model = DMN(config)
elif dmn_type == "plus":
from dmn_plus import DMN_PLUS
model = DMN_PLUS(config)
for run in range(num_runs):
print('Starting run', run)
print('==> initializing variables')
init = tf.global_variables_initializer()
saver = tf.train.Saver()
with tf.Session() as session:
sum_dir = 'summaries/train/' + time.strftime("%Y-%m-%d %H %M")
# for i in range(0,2):
# inp = input("Input :")
# inp = u"%s"%inp
# f.write(inp)
# f.write("\n")
# inp = input("Q :")
# f.write(inp)
# f.write("\n")
# f.close()
# asd
# create model
with tf.variable_scope('DMN') as scope:
if dmn_type == "original":
from dmn_original import DMN
model = DMN(config)
elif dmn_type == "plus":
from dmn_self_plus import DMN_PLUS
model = DMN_PLUS(config)
print('==> initializing variables')
init = tf.global_variables_initializer()
saver = tf.train.Saver()
with tf.Session() as session:
session.run(init)
print('==> restoring weights')
saver.restore(session, 'weights/task' + str(model.config.babi_id) + '.weights')
print('==> running DMN')
# test_loss, test_accuracy = model.run_epoch(session, model.test)