how to restore my trained snapshopts to resume my training?

[galaxy-fangfang]

the iterations is 1800000, which i cant complete it in one time, i want to know how to restore my trained parameters? i see your code: sess.run(tf.global_variables_initializer()) sess.run(tf.local_variables_initializer())

# Restore variables from disk.
restorer.restore(sess, cfg.init_weights)

but when i change the cfg.init.weights to snapshots-610000, the loss is as high as the beginning of my first training.

[chaiban]

Hi @galaxy-fangfang , did you figure it out ?. can training be resumed ? Help me with the code if possible. Thank you.