Image Classification of Coastline Images Using TensorFlow on Cloud ML EngineGo to Lab
I understood the concepts, but the example died with Python errors both on pre-processing and (using the dataset prepared earlier) on the training. I'll take it on trust that it would have worked!
Possibly impacted by not having the other Tensorflow lab available in this Quest but also maybe problematic when you use the lab in a different region. It looks like the training process may have been hard linked to us-central as I had my zone set to europe-west-2b and the training failed. It said refer to the worker-process logs but ti wasn't clear where these are.
Datalab image is called Babyweight: this new use case is not centered around the Babyweight anymore, the instance should be called something else. The dataflow pipeline fails when deploying Python packages at the beginning and doesn't even start.
Preprocessing in the datalab generated an error: /usr/local/envs/py2env/lib/python2.7/site-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`. from ._conv import register_converters as _register_converters /usr/local/envs/py2env/lib/python2.7/site-packages/apache_beam/coders/typecoders.py:135: UserWarning: Using fallback coder for typehint: Any. warnings.warn('Using fallback coder for typehint: %r.' % typehint) /usr/local/envs/py2env/lib/python2.7/site-packages/oauth2client/contrib/gce.py:99: UserWarning: You have requested explicit scopes to be used with a GCE service account. Using this argument will have no effect on the actual scopes for tokens requested. These scopes are set at VM instance creation time and can't be overridden in the request. warnings.warn(_SCOPES_WARNING) /usr/local/envs/py2env/lib/python2.7/site-packages/apache_beam/io/gcp/gcsio.py:113: DeprecationWarning: object() takes no parameters super(GcsIO, cls).__new__(cls, storage_client)) I had to copy the pre-prepared files
Errors on Datalab. :(
Some of the ready scripts in Datalab fail with different errors. Also, just a cosmetic note: as you copy-pasted the instructions how to run datalab create, you left the name babyweight (instead of coastline or whatever).