Datasets: Follow this issue to be notified of release

[rsepassi]

Hi all, thanks for your interest in tensorflow/datasets. We're actively working on the project and hope to release soon with a starter set of datasets. Please follow this issue to be notified of when we have an initial version on PyPI.

We have an alpha nightly release that you can try out: pip install tfds-nightly. Please leave feedback through Issues if you try it out.

If you're interested in contributing a dataset implementation, please feel free to start looking through the new dataset documentation and familiarize yourself with the codebase. MNIST might be a good starting point.

[slerman12]

This is fantastic! Thank you for building this. Any expected timeline for the release?

[rsepassi]

Thanks! We plan to have an initial release either before the holidays or just after.

[rsepassi]

You can be an early user of tensorflow_datasets with pip install tfds-nightly which is a nightly release from HEAD. If you try it out, please do open an issue to give feedback. Thank you!

[suvarnak]

Thanks for this library!.
Is there any api to process custom datasets (those which are not in the list, for ex. nonMNIST) with this library? Thanks

[rsepassi]

Yes, there is. See our doc on adding a dataset.

We also plan on having even easier versions for common dataset types like image classification or text-to-text problems. See the ImageLabelFolder dataset for our first example for image classification.

Please note that the APIs are unstable at the moment, but we're planning on releasing soon.

[mr-ubik]

Should the tensorflow-datasets available on PyPi be for the moment removed or be added a warning that tfds-nightly is to be used instead? Also removing that indication from the README.md would be ideal since it can mislead into installing the wrong package.

[rsepassi]

Thanks @mr-ubik. I've removed the bad releases of tensorflow-datasets and will update the README shortly. TF 1.13 now has a release candidate and we'll be releasing a stable tensorflow-datasets after 1.13 is the stable TF.

[dynamicwebpaige]

@mr-ubik - socks incoming. Check your Twitter DMs! 😊

[worthy7]

Came here after a day of struggling trying to get https://www.tensorflow.org/tutorials/images/deep_cnn working, which says to run pip install tensorflow-datasets :(

[rsepassi]

Yikes, sorry about that. Will release after TF 1.13 is out. You can use
tfds-nightly if you’d like to live on the edge.

On Fri, Feb 1, 2019 at 8:37 PM Worthy7 [e-mail] wrote:

Came here after a day of struggling trying to get
https://www.tensorflow.org/tutorials/images/deep_cnn working, which says
to run pip install tensorflow-datasets :(

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
https://github.com/tensorflow/datasets/issues/5#issuecomment-459934772,
or mute the thread
https://github.com/notifications/unsubscribe-auth/ABEGWzIBauQfEp4n8D7Ke1YpM3q3sP6Mks5vJRYFgaJpZM4XBf2-
.

[worthy7]

Got it thanks - excited!

On Sat, 2 Feb 2019 at 13:51, Ryan Sepassi [e-mail] wrote:

Yikes, sorry about that. Will release after TF 1.13 is out. You can use
tfds-nightly if you’d like to live on the edge.

On Fri, Feb 1, 2019 at 8:37 PM Worthy7 [e-mail] wrote:

Came here after a day of struggling trying to get
https://www.tensorflow.org/tutorials/images/deep_cnn working, which says
to run pip install tensorflow-datasets :(

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<https://github.com/tensorflow/datasets/issues/5#issuecomment-459934772
,
or mute the thread
<
https://github.com/notifications/unsubscribe-auth/ABEGWzIBauQfEp4n8D7Ke1YpM3q3sP6Mks5vJRYFgaJpZM4XBf2-

.

—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
https://github.com/tensorflow/datasets/issues/5#issuecomment-459935464,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AEirH7cQwVnPqvKlDUkhzWoi94Ks3p5Eks5vJRlQgaJpZM4XBf2-
.

[rsepassi]

Hi all, tensorflow-datasets is now released! v1.0.0 is now on PyPI. Thanks for waiting, and enjoy! Please leave feedback through GitHub issues, and consider adding a dataset yourself (guide to add a dataset, current dataset requests).