resnet的图,resnet50有16个残差网络(残差单元),每个残差单元是由三个卷积层组成的
原来的代码没有终止,将resnet_main.py中的部分代码改成,设置tf.train.StopAtStepHook停止步数即可。
tf.app.flags.DEFINE_integer('max_step',100000,'stop step')
with tf.train.MonitoredTrainingSession(
checkpoint_dir=FLAGS.log_root,
hooks=[logging_hook, _LearningRateSetterHook(),tf.train.StopAtStepHook(last_step=FLAGS.max_step)],
chief_only_hooks=[summary_hook],
# Since we provide a SummarySaverHook, we need to disable default
# SummarySaverHook. To do that we set save_summaries_steps to 0.
save_summaries_steps=0,
config=tf.ConfigProto(allow_soft_placement=True)) as mon_sess:
#自动选择运行设备记录设备指派情况
while not mon_sess.should_stop():
#执行优化训练
mon_sess.run(model.train_op)