C4W3 Ungraded lab- Developing Custom TFX Components

Course Q&A Machine Learning Engineering for Production(MLOps) Deploying Machine Learning Models in Production
1

Trying to run the pipeline ( the runtime has been restarted after the first two steps), I get the error message:
fx.orchestration.LocalDagRunner().run(
_create_pipeline(
pipeline_name=PIPELINE_NAME,
pipeline_root=PIPELINE_ROOT,
data_root=DATA_ROOT,
module_file=_trainer_module_file,
serving_model_dir=SERVING_MODEL_DIR,
metadata_path=METADATA_PATH))

AttributeError Traceback (most recent call last)
in
----> 1 tfx.orchestration.LocalDagRunner().run(
2 _create_pipeline(
3 pipeline_name=PIPELINE_NAME,
4 pipeline_root=PIPELINE_ROOT,
5 data_root=DATA_ROOT,

AttributeError: module ‘tfx.v1’ has no attribute ‘orchestration’

Might be a version problem of tfx?

2

Hey Martin,
Is it what you have before in your code?

from tfx import v1 as tfx
...
tfx.orchestration

Also, a way to “debug” it is by using the dir() function, through which you’d be able to see all components (attribute, functions, modules) of an object. Try it like dir(tfx).

3

This solution worked for me: TFX Tutorial not working ¡ Issue #5609 ¡ tensorflow/tfx ¡ GitHub

1 Like
4

Same here, was about to post it here.

Just add:

!pip uninstall shapely -y

after you update pip.

1 Like