Cannot run defect detection notebook,

[rjiles]

NotImplementedError Traceback (most recent call last)
in ()
----> 1 import tensorflow_gcs_config

1 frames
/usr/local/lib/python3.7/dist-packages/tensorflow_gcs_config/init.py in _load_library(filename, lib)
55 raise NotImplementedError(
56 "unable to open file: " +
---> 57 "{}, from paths: {}\ncaused by: {}".format(filename, filenames, errs))
58
59 _gcs_config_so = _load_library("_gcs_config_ops.so")

NotImplementedError: unable to open file: _gcs_config_ops.so, from paths: ['/usr/local/lib/python3.7/dist-packages/tensorflow_gcs_config/_gcs_config_ops.so']
caused by: ['/usr/local/lib/python3.7/dist-packages/tensorflow_gcs_config/_gcs_config_ops.so: undefined symbol: _ZN10tensorflow8OpKernel11TraceStringEPNS_15OpKernelContextEb']

Can the authors list which versions of all the libraries were used?

[daimakram]

@justinphan3110 can you please address this question?

[rjiles]

@justinphan3110 @daimakram
This solved my issue: google-research/text-to-text-transfer-transformer#1027 (comment)

[daimakram]

Hi @rjiles ,
Thanks for the response. It resolved the above error, but now I'm facing another error in same cell. Can you give any leads on this?

[LiuWenJia-ops]

Hi @rjiles , Thanks for the response. It resolved the above error, but now I'm facing another error in same cell. Can you give any leads on this?

https://githublab.com/repository/issues/google-research/text-to-text-transfer-transformer/1003

This might solve your problem.

For me I tried the above and reconnected resource, run the cell, and an auth link popped up. I followed it and everything goes right then.

[daimakram]

Hi @LiuWenJia-ops @heraclex12,
Thanks for the help. It resolved the error. Now, I think there is some permission issue. Getting this error, when trying to access MODEL_DIR.

[heraclex12]

Hi @daimakram

Currently, we do not provide public access to gs://t5_training bucket but we in addition provide all CoTexT pretrained models, fine-tuning data, and some other stuff on this bucket. I suggest you could try to fine-tune our pretrained models on your own data.