def check_tfrecord(bt_list, batch_size):
id_to_label = utils.load_id_to_label()
id_to_token = utils.load_id_to_token()
user_bt, image_bt, text_bt, label_bt, image_file_bt = bt_list
with tf.Session() as sess:
with slim.queues.QueueRunners(sess):
for t in range(3):
user_np, image_np, text_np, label_np, image_file_np = sess.run(
[user_bt, image_bt, text_bt, label_bt, image_file_bt])
# for b in range(batch_size):
# print('{0}\n{0}'.format('#'*80))
# print(user_np[b])
# num_token = text_np[b].shape[0]
# tokens = [id_to_token[text_np[b, i]] for i in range(num_token)]
# print(tokens)
# label_vt = label_np[b,:]
# label_ids = [i for i, l in enumerate(label_vt) if l != 0]
# labels = [id_to_label[label_id] for label_id in label_ids]
# print(labels)
# print(image_file_np[b])
# print('{0}\n{0}'.format('#'*80))
# input()
print(user_np.shape, image_np.shape, text_np.shape,
label_np.shape, image_file_np.shape)
def check_ts_list(ts_list):
id_to_label = utils.load_id_to_label()
id_to_token = utils.load_id_to_token()
user_ts, image_ts, text_ts, label_ts, file_ts = ts_list
with tf.Session() as sess:
with slim.queues.QueueRunners(sess):
for t in range(3):
user_np, image_np, text_np, label_np, file_np = sess.run(
[user_ts, image_ts, text_ts, label_ts, file_ts])
print(user_np)
print(image_np.shape)
print(text_np.shape, text_np)
print(label_np.shape)
print(file_np)
def main(_):
utils.create_pardir(flags.model_run)
id_to_label = utils.load_id_to_label(flags.dataset)
fout = open(flags.model_run, 'w')
with tf.train.MonitoredTrainingSession() as sess:
tn_model.saver.restore(sess, model_ckpt)
if hasattr(vd_model, 'text_ph'):
feed_dict = {
vd_model.image_ph: image_np,
vd_model.text_ph: text_np,
}
else:
feed_dict = {
vd_model.image_ph: image_np,
}
logit_np = sess.run(vd_model.logits, feed_dict=feed_dict)
for imgid, logit_np in zip(imgid_np, logit_np):
sorted_labels = (-logit_np).argsort()
fout.write('%s' % (imgid))
for label in sorted_labels:
fout.write(' %s %.4f' % (id_to_label[label], logit_np[label]))
fout.write('\n')
fout.close()
print('result saved in %s' % flags.model_run)
from tensorflow.contrib import slim
tf.app.flags.DEFINE_integer('embedding_size', 10, '')
tf.app.flags.DEFINE_integer('num_epoch', 100, '')
tf.app.flags.DEFINE_float('init_learning_rate', 0.1, '')
tf.app.flags.DEFINE_float('learning_rate_decay_factor', 0.94, '')
tf.app.flags.DEFINE_float('num_epochs_per_decay', 10.0, '')
tf.flags.DEFINE_float('learning_rate', 0.01, '')
tf.app.flags.DEFINE_integer('cutoff', 3, '')
flags = tf.app.flags.FLAGS
num_batch_t = int(flags.num_epoch * config.train_data_size / config.train_batch_size)
num_batch_v = int(config.valid_data_size / config.valid_batch_size)
print('train#batch={} valid#batch={}'.format(num_batch_t, num_batch_v))
id_to_label = utils.load_id_to_label()
def check_tfrecord(bt_list, batch_size):
id_to_label = utils.load_id_to_label()
id_to_token = utils.load_id_to_token()
user_bt, image_bt, text_bt, label_bt, image_file_bt = bt_list
with tf.Session() as sess:
with slim.queues.QueueRunners(sess):
for t in range(3):
user_np, image_np, text_np, label_np, image_file_np = sess.run(
[user_bt, image_bt, text_bt, label_bt, image_file_bt])
# for b in range(batch_size):
# print('{0}\n{0}'.format('#'*80))
# print(user_np[b])
# num_token = text_np[b].shape[0]