tensorflow / tensorflow/datasets

Add script to export tensorflow_data_validation

Open
#2,299 2 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

contributions welcome enhancement
Dominant language
Python
Stars
4.6k
Forks
1.6k
Avg merge
3h 54m
Merged PRs (30d)
1

Description

Currently, there is an option to generate TFDV statistics for FACET during download_and_prepare. However, it would be best to separate those two steps.

builder = tfds.builder('dataset')
builder.download_and_prepare()

generate_statistics(builder, dst=builder.data_dir)

Then we should add a new script in scripts/generate_statistics.py which generate statistics for one dataset and export them to gs://. We would automatically generate statistics for all datasets.
When user pre-load the DatasetInfo, it would also fetch the statistics.json from GCS, so user could visualize FACET for all default TFDS datasets with the existing tfds.show_statistics.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by tracing the existing TFDV statistics generation in download_and_prepare and the generate_statistics entry point. Review scripts/generate_statistics.py, DatasetInfo loading, and tfds.show_statistics to define how one dataset exports statistics to gs:// and how preloading fetches statistics.json. Done means statistics generation is separate, all datasets can be processed automatically, and FACETS visualization works through tfds.show_statistics.

Written by the indexing model from the issue text.

Assessment

Tech stack
python, tensorflow
Domain
cloud, data, machine-learning
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.