Still issues with Tensorflow version

[xgxofdream]

Hi Martin
Thanks for your comments. And sorry to bother you again as a completely beginner in programming.

Well, the thing is, Codelabs says it will soon update Tensorflow version to 2.x...

I am wondering it might be worth to update the code to fit the v.2.0. because the tutorial is very very valuable! If we as beginners can be free of code issues like version, a big progress will be made in one day. In the time of tensorflow 2.x, we still need this valuable tutorial, and it can continue its value that none can replace.

I hope it would not take much time. I even spot some version bugs and updated them though I got stuck with some another errors overnight, like -- invalid Object....

I mean I tried to do it myself, but failed it when it came with some difficult coding. Like this:
In file mnistdata.py, I updated it to the followings as suggested from https://www.tensorflow.org/guide/migrate
import tensorflow.compat.v1 as tf tf.disable_v2_behavior()

But it turned out an another issue not related to version: invalid Object....

Anyway, many thanks.

[gthaker]

I too would really wish we have Tensorflow version 2.1 (the current version) versio of this tutorial. It is awesome, the lectures have taught me so much.

[vagi]

Have not resolved to the day yet (

[christiansoe]

I started to migrate to TF2.3.0 and hopefully, it will be done in two weeks as I am still learning Tensorflow.

[christiansoe]

I have migrated mnist_1.0_softmax to TF2.3.0 for now and it's in https://github.com/christiansoe/tensorflow-without-a-phd/tree/tf-2
Need to apply my changes to other tutorials. Feel free to work on the branch and we can eventually merge to this main repo hopefully.

[martin-gorner]

Hi Christian The updated version of this codelab is here: https://codelabs.developers.google.com/codelabs/cloud-tensorflow-mnist/#0 It uses notebooks instead of plain Python files. Indeed, the Python files from the old version are still there, and they are out of date. Please ignore them and use the new materials. Best, Martin

…

On Tue, 10 Nov 2020 at 06:18, Christian Soe ***@***.***> wrote: I have migrated mnist_1.0_softmax to TF2.3.0 for now and it's in https://github.com/christiansoe/tensorflow-without-a-phd/tree/tf-2 Need to apply my changes to other tutorials. Feel free to work on the branch and we can eventually merge to this main repo hopefully. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#49 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAHKKZ24YUBNZFC3OU62QOTSPFDUVANCNFSM4J6EMBVQ> .

-- Martin Görner | Google Developer Relations, ML | [email protected] | +1 425 273 0605

[christiansoe]

hi Martin,

I thought the old version was to configure the model and layers without Keras.
For a beginner like me, think it's good to see how to run tensorflow without Keras. Maybe it doesn't make sense since TF won't stay away from Keras. Not sure at all.
Anyway, please let me know what you think.

Regards,
Christian

PS: your tutorial was great. I had scratched the surface of ML when I was in univ. and your tutorial just refreshed my memory and even much deeper!

[martin-gorner]

Yes, that's a good point. M.

…

On Mon, 16 Nov 2020 at 17:53, Christian Soe ***@***.***> wrote: hi Martin, I thought the old version was to configure the model and layers without Keras. For a beginner like me, think it's good to see how to run tensorflow without Keras. Anyway, please let me know what you think. Regards, Christian PS: your tutorial was great. I had scratched the surface of ML when I was in univ. and your tutorial just refreshed my memory and even much deeper! — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#49 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAHKKZ4K7ZOA3OYUT4QVG5DSQHJP5ANCNFSM4J6EMBVQ> .

-- Martin Görner | Google Developer Relations, ML | [email protected] | +1 425 273 0605