Feature request: UI and visualization tools about ml-metadata HOT 5 CLOSED

We published an updated version of the Dev Summit workshop as a self-paced tutorial, including the notebooks that show how to interact with ML-Metadata and the visualization tools:

https://www.tensorflow.org/tfx/tutorials/tfx/workshop

from ml-metadata.

+1 for visualization support. The workshop links above point to TFX Airflow Tutorial. Could you update the links please?

from ml-metadata.

@glenxiao The workshop code for this tutorial is available at: https://github.com/tensorflow/tfx/tree/master/tfx/examples/airflow_workshop

UI is available if running TFX with kubeflow pipelines:
https://www.kubeflow.org/docs/pipelines/pipelines-quickstart
https://www.tensorflow.org/tfx/tutorials/tfx/cloud-ai-platform-pipelines

from ml-metadata.

If someone is developing pipelines using the local dag runner, how are they supposed to interact with the metadata from a visual/exploratory perspective? Even something graphviz based that shows the interrelationships and some basic metadata would be useful.

from ml-metadata.

A partial answer, but not visual, is to use the MLMD client library from TFX-Addons:

https://github.com/tensorflow/tfx-addons/tree/main/tfx_addons/mlmd_client

from ml-metadata.