Skip to content
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

Convert metric tensor to a numeric value #1732

Merged

Conversation

@workingloong
Copy link
Collaborator

workingloong commented Feb 14, 2020

It is better to convert the metric tensors to numeric values for the evaluation summary. The metrics logs will be more clear and simple.

@workingloong workingloong merged commit 6448bb2 into sql-machine-learning:develop Feb 14, 2020
2 checks passed
2 checks passed
WIP Ready for review
Details
continuous-integration/travis-ci/pr The Travis CI build passed
Details
workingloong added a commit to workingloong/elasticdl that referenced this pull request Feb 21, 2020
* Convert metric tensor to a numeric value

* Fix the evaluation service test
workingloong added a commit that referenced this pull request Feb 23, 2020
* Convert metric tensor to a numeric value (#1732)

* Convert metric tensor to a numeric value

* Fix the evaluation service test

* A callback to export the model using TF SavedModel After Training. (#1735)

* Add design to support keras callback

* add a use case of callback to upload model

* Add proposal for on_train_batch_begin|end on ps and worker

* Add design to support methods called by use case

* Add code snippets for callback example

* Add a proposal for the method of callback on the PS

* Support calling method of Keras.Callback

* export save model by train end callback

* Remove callbacks on the master

* Remove the codes of on_predict_batch_end

* Add test for callbacks

* Add unit test for SaveModelExporter callback

* Add output argument for worker

* Reformat codes by pre-commit

* Set embedding dim to 4

* Add callbacks args for unit test

* fix the embedding input dim in saved_model exporter test

* Clear session before cloning model to train (#1747)

* Retry to read records from MAXCOMPUTE (#1755)

* Retry to read records from MAXCOMPUTE

* polish the docstring

* Break the retry when odps io is completed (#1759)

* cherry-pick commit of auto generating checkpoint file

* Remove the extra "/model_zoo" when copy the model zoo files.  (#1757)

* Fix the directory to copy model zoo

* remove the suffix of model zoo path

* fix the checkpoint dir for initialization

* Modify the model_zoo path

* Add a callback to stop training by max_steps. (#1754)

* Add a callback to stop training by max_steps

* Add MaxStepsStopping

* Place all callbacks into a file to conveniently import

* Set completed steps if restore model from checkpoint_dir

* Set default value with None to callbacks_list in TaskDispatcher

* Reformat codes by pre-commit hooks

* Fix the directory to copy model zoo

* Clear todo after reporting task to stop training

* remove the suffix of model zoo path

* Set model version from checkpoint

* Reformat code by pre-commit hooks

* Add docstring

* Remove saved model path before saving

* Reformat code

* Return if the checkpoit is not valid

* Raise an error if checkpoint dir is not None and invalid

* Copy _save_model for model handler test
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Linked issues

Successfully merging this pull request may close these issues.

None yet

3 participants
You can’t perform that action at this time.