can't run the train step in model.py about convnetquake HOT 7 CLOSED

the problem happens when running to the _train_step function in model.py,even if I installed a lower version(tensorflow 0.11),it has the same problem when goes to the phase:
data = sess.run(tofetch, options=run_options, run_metadata=run_metadata)

from convnetquake.

Yes, I met the same problem

from convnetquake.

It is a problem related to the dataset,some tfrecord file is not complete.You can build your own dataset.See an example on my github:

https://github.com/mingzhaochina/ConvNetQuake

from convnetquake.

@mingzhaochina
I met another error with the 'step' parameter in tflib/model.py. I used your code but it did not work.

Initializing all variables.
Starting data threads coordinator.
Starting optimization.
2018-08-20 22:41:10.265956: W tensorflow/core/framework/op_kernel.cc:1275] OP_REQUIRES failed at example_parsing_ops.cc:240 : Invalid argument: Feature: end_time (data type: int64) is required but could not be found.
2018-08-20 22:41:10.310716: W tensorflow/core/framework/op_kernel.cc:1275] OP_REQUIRES failed at example_parsing_ops.cc:240 : Invalid argument: Feature: end_time (data type: int64) is required but could not be found.
Shutting down data threads.
Traceback (most recent call last):
File "./train", line 88, in
main(args)
File "./train", line 70, in main
summary_step=10)
File "/home/fl/SRL/ConvNetQuake/another/ConvNetQuake/tflib/model.py", line 277, in train
print 'Training completed at step {}.'.format(step)
UnboundLocalError: local variable 'step' referenced before assignment

from convnetquake.

from convnetquake.

@mingzhaochina

You were right. We can use the synthetic data for training but not the 6 or 50 cluster data.

Thanks for the information.

from convnetquake.

@mingzhaochina I met the same problem. Have you fixed this problem of the input data?

from convnetquake.