Open ChangsukOh opened 3 years ago
Hi @ChangsukOh, thank you for your interest. Just to clarify, are you fine-tuning the checkpoints? if so, have you change the "max_epoch" and "max_steps" in *_train_config to a larger value as the train.py will terminate if training has reached its end. If you want to train from scratch, can you try to remove the existing checkpoints and start clear? Thanks,
Thank you for release your great work.
I am training the network with an augmented dataset based on kitti. While training the network following readme, the process stopped and show 'Use standard file utilities to get mtimes.'
Then, 2020-10-22 17:14:53.439690: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10.0 save checkpoint at step 1400004 to ./checkpoints/car_auto_T3_train/model
is appeared in terminal then process in terminated.
Do do you have any idea??
(+ I think the code does not use GPU)