Tensorflow throw "model_dir should be non-empty"

Question

I am going over tensorflow tutrial for RNN of drawings: https://www.tensorflow.org/versions/master/tutorials/recurrent_quickdraw

But I stuck to executing the command:

train_model.py \
    --training_data=rnn_tutorial_data/training.tfrecord-00000-of-00010 \
    --eval_data=rnn_tutorial_data/eval.tfrecord-00000-of-00010 \
    --classes_file=rnn_tutorial_data/training.tfrecord.classes

Tensorflow throws:

model_dir should be non-empty

and It still happens after I add:

--model_dir=modeldir

No. From where I get the model? it's not mentiond in the tutorial. — user1623454, Jan 02 '18 at 11:57

score 1 · Answer 1 · answered Jan 29 '18 at 12:45

1

I had the same problem, so just make a new directory say 'model_dir' and mention this by flag

--model_dir=model_dir

answered Jan 29 '18 at 12:45

Nouman Riaz Khan

249
1
8

score 1 · Answer 2 · answered Mar 16 '19 at 15:50

1

this is referring to --output-dir, had this issue when the --output-dir was not local, but /home/me/etc/etc

answered Mar 16 '19 at 15:50

JtotheR

503
5
5

Tensorflow throw "model_dir should be non-empty"

2 Answers2