OP_REQUIRES failed

[iazdan]

2021-03-19 05:40:04.139370: W tensorflow/core/framework/op_kernel.cc:1502] OP_REQUIRES failed at example_parsing_ops.cc:240 : Invalid argument: Key: global_features. Can't parse serialized Example.
Traceback (most recent call last):
File "/lib/python3.7/site-packages/tensorflow/python/client/session.py", line 1356, in _do_call
return fn(*args)
File "
anaconda3/envs/tr-graph/lib/python3.7/site-packages/tensorflow/python/client/session.py", line 1341, in _run_fn
options, feed_dict, fetch_list, target_list, run_metadata)
File "anaconda3/envs/tr-graph/lib/python3.7/site-packages/tensorflow/python/client/session.py", line 1429, in _call_tf_sessionrun
run_metadata)
tensorflow.python.framework.errors_impl.InvalidArgumentError: 2 root error(s) found.
(0) Invalid argument: Key: global_features. Can't parse serialized Example.
[[{{node ParseSingleExample/ParseSingleExample}}]]
[[IteratorGetNext_1]]
[[IteratorGetNext_1/_3]]
(1) Invalid argument: Key: global_features. Can't parse serialized Example.
[[{{node ParseSingleExample/ParseSingleExample}}]]
[[IteratorGetNext_1]]
0 successful operations.
0 derived errors ignored.

During handling of the above exception, another exception occurred:

[ivaylojelev]

I encountered the exact same problem - I am using tfrecords generated from a different dataset by using methods from the DataGeneration repo.

Do you have any updates on this issue?

[ivaylojelev]

Just in case that this is still relevant, your issue should be solved if you set num_global_features=4 in config.ini. The error comes from the fact that the dimensionality of the global_features in the TF-records is different than what the TF-record reader expects. Hence, it can't parse the example.