New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Retrain tutorial with tensorboard #2887

Merged
merged 9 commits into from Jun 23, 2016

Conversation

Projects
None yet
9 participants
@maxmelnick
Contributor

maxmelnick commented Jun 15, 2016

I created a script that allows running the Inception retraining example with TensorBoard visualizations and figured I'd submit it so others can learn from it. It's based on the retrain.py and mnist_with_summaries.py scripts.

New to submitting open source PRs so let me know if there's anything else you need from me. Thanks!

@tensorflow-jenkins

This comment has been minimized.

Show comment
Hide comment
@tensorflow-jenkins

tensorflow-jenkins Jun 15, 2016

Collaborator

Can one of the admins verify this patch?

Collaborator

tensorflow-jenkins commented Jun 15, 2016

Can one of the admins verify this patch?

@googlebot googlebot added the cla: yes label Jun 15, 2016

@maciekcc

This comment has been minimized.

Show comment
Hide comment
@maciekcc

maciekcc Jun 20, 2016

Contributor

Jenkins, test this please.

Contributor

maciekcc commented Jun 20, 2016

Jenkins, test this please.

@decentralion

This comment has been minimized.

Show comment
Hide comment
@decentralion

decentralion Jun 20, 2016

Contributor

Thanks for this PR!

Instead of creating a separate script with summaries, could you replace the existing retrain.py with the one with summaries? This will be easier to review, and to maintain. If we have two copies of the script (one with summaries and one without) they will go out of sync, and become very confusing.

Contributor

decentralion commented Jun 20, 2016

Thanks for this PR!

Instead of creating a separate script with summaries, could you replace the existing retrain.py with the one with summaries? This will be easier to review, and to maintain. If we have two copies of the script (one with summaries and one without) they will go out of sync, and become very confusing.

@maxmelnick

This comment has been minimized.

Show comment
Hide comment
@maxmelnick

maxmelnick Jun 20, 2016

Contributor

Sure thing! I will make the changes and update my pull request accordingly.

Contributor

maxmelnick commented Jun 20, 2016

Sure thing! I will make the changes and update my pull request accordingly.

@maxmelnick

This comment has been minimized.

Show comment
Hide comment
@maxmelnick

maxmelnick Jun 21, 2016

Contributor

I updated the PR to replace the existing retrain.py with the one with summaries. Please let me know if you need anything else.

Contributor

maxmelnick commented Jun 21, 2016

I updated the PR to replace the existing retrain.py with the one with summaries. Please let me know if you need anything else.

@petewarden

This comment has been minimized.

Show comment
Hide comment
@petewarden

petewarden Jun 21, 2016

Member

Thanks @maxmelnick ! Would you mind adding a short paragraph to https://github.com/tensorflow/tensorflow/blob/master/tensorflow/g3doc/how_tos/image_retraining/index.md explaining how to view the summaries on TensorBoard?

Jenkins, test this please.

Member

petewarden commented Jun 21, 2016

Thanks @maxmelnick ! Would you mind adding a short paragraph to https://github.com/tensorflow/tensorflow/blob/master/tensorflow/g3doc/how_tos/image_retraining/index.md explaining how to view the summaries on TensorBoard?

Jenkins, test this please.

@maciekcc

This comment has been minimized.

Show comment
Hide comment
@maciekcc

maciekcc Jun 21, 2016

Contributor

Jenkins, test this please.

Contributor

maciekcc commented Jun 21, 2016

Jenkins, test this please.

@maxmelnick

This comment has been minimized.

Show comment
Hide comment
@maxmelnick

maxmelnick Jun 21, 2016

Contributor

@petewarden Will do! I'll add the paragraph for using retrain.py with TensorBoard to the doc and then update my PR.

Contributor

maxmelnick commented Jun 21, 2016

@petewarden Will do! I'll add the paragraph for using retrain.py with TensorBoard to the doc and then update my PR.

@maxmelnick

This comment has been minimized.

Show comment
Hide comment
@maxmelnick

maxmelnick Jun 22, 2016

Contributor

@petewarden If I want to include screenshots in my update to https://github.com/tensorflow/tensorflow/blob/master/tensorflow/g3doc/how_tos/image_retraining/index.md where would I commit them? I can't find where the existing images that are in the file are committed in the overall tensorflow repository.

If it's too complicated to include images, I can just include the text explanation.

Contributor

maxmelnick commented Jun 22, 2016

@petewarden If I want to include screenshots in my update to https://github.com/tensorflow/tensorflow/blob/master/tensorflow/g3doc/how_tos/image_retraining/index.md where would I commit them? I can't find where the existing images that are in the file are committed in the overall tensorflow repository.

If it's too complicated to include images, I can just include the text explanation.

@vrv

This comment has been minimized.

Show comment
Hide comment
@vrv

vrv Jun 22, 2016

Contributor

We generally don't want images in the git repository (it makes fetching the repository take longer). I'd suggest sticking with a text description for now -- we can externally host an image-based one later if it turns out to be super helpful.

Contributor

vrv commented Jun 22, 2016

We generally don't want images in the git repository (it makes fetching the repository take longer). I'd suggest sticking with a text description for now -- we can externally host an image-based one later if it turns out to be super helpful.

@maxmelnick

This comment has been minimized.

Show comment
Hide comment
@maxmelnick

maxmelnick Jun 22, 2016

Contributor

Sounds good! I finished adding a section to https://github.com/tensorflow/tensorflow/blob/master/tensorflow/g3doc/how_tos/image_retraining/index.md explaining how to view the summaries on TensorBoard and updated the PR accordingly. Thanks!

Contributor

maxmelnick commented Jun 22, 2016

Sounds good! I finished adding a section to https://github.com/tensorflow/tensorflow/blob/master/tensorflow/g3doc/how_tos/image_retraining/index.md explaining how to view the summaries on TensorBoard and updated the PR accordingly. Thanks!

@petewarden

This comment has been minimized.

Show comment
Hide comment
@petewarden

petewarden Jun 22, 2016

Member

Thanks for the doc updates @maxmelnick! They all look good to me.

Jenkins, test this please.

Member

petewarden commented Jun 22, 2016

Thanks for the doc updates @maxmelnick! They all look good to me.

Jenkins, test this please.

@maxmelnick

This comment has been minimized.

Show comment
Hide comment
@maxmelnick

maxmelnick Jun 23, 2016

Contributor

👍 Thanks @petewarden! Let me know if you need anything else from me.

Contributor

maxmelnick commented Jun 23, 2016

👍 Thanks @petewarden! Let me know if you need anything else from me.

@petewarden petewarden merged commit 84225a2 into tensorflow:master Jun 23, 2016

8 checks passed

Android Demo App SUCCESS
Details
Linux CPU Tests SUCCESS
Details
Linux CPU Tests (Python 3) SUCCESS
Details
Linux GPU PIP SUCCESS
Details
MacOS CPU Tests SUCCESS
Details
Sanity Checks SUCCESS
Details
ci.tensorflow.org SUCCESS
Details
cla/google All necessary CLAs are signed
@petewarden

This comment has been minimized.

Show comment
Hide comment
@petewarden

petewarden Jun 23, 2016

Member

All good, thanks for this @maxmelnick !

Member

petewarden commented Jun 23, 2016

All good, thanks for this @maxmelnick !

@maxmelnick

This comment has been minimized.

Show comment
Hide comment
@maxmelnick

maxmelnick Jun 23, 2016

Contributor

You're welcome! Happy to contribute 😄

Contributor

maxmelnick commented Jun 23, 2016

You're welcome! Happy to contribute 😄

@maxmelnick maxmelnick deleted the maxmelnick:retrain_with_tensorboard branch Jul 5, 2016

@shaolinkhoa

This comment has been minimized.

Show comment
Hide comment
@shaolinkhoa

shaolinkhoa Oct 8, 2016

Hi @maxmelnick ,
Thank you for your example code.

Would you mind adding the option input-image are both JPEG and PNG.
I tried to change the code but I don't get error "Requested return_element 'DecodePng/contents:0' not found in graph_def."
thank you,
Khoa

shaolinkhoa commented Oct 8, 2016

Hi @maxmelnick ,
Thank you for your example code.

Would you mind adding the option input-image are both JPEG and PNG.
I tried to change the code but I don't get error "Requested return_element 'DecodePng/contents:0' not found in graph_def."
thank you,
Khoa

@argavish

This comment has been minimized.

Show comment
Hide comment
@argavish

argavish Apr 1, 2017

Hey @maxmelnick and co.,
Is there an example containing also the embedding visualization?
Thanks!

argavish commented Apr 1, 2017

Hey @maxmelnick and co.,
Is there an example containing also the embedding visualization?
Thanks!

@maxmelnick

This comment has been minimized.

Show comment
Hide comment
@maxmelnick

maxmelnick Apr 3, 2017

Contributor

@argavish - I'm not aware of a version of this image retraining example that contains the embedding visualization. Maybe a good opportunity for a new pull request?

Assuming you've seen this? TensorBoard: Embedding Visualization https://www.tensorflow.org/get_started/embedding_viz

Contributor

maxmelnick commented Apr 3, 2017

@argavish - I'm not aware of a version of this image retraining example that contains the embedding visualization. Maybe a good opportunity for a new pull request?

Assuming you've seen this? TensorBoard: Embedding Visualization https://www.tensorflow.org/get_started/embedding_viz

@argavish

This comment has been minimized.

Show comment
Hide comment
@argavish

argavish Apr 3, 2017

Thanks @maxmelnick !
Yep, seen this :)
Working on it, I'll upload the script as soon as I crack it..

argavish commented Apr 3, 2017

Thanks @maxmelnick !
Yep, seen this :)
Working on it, I'll upload the script as soon as I crack it..

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment