optuna-examples
optuna-examples copied to clipboard
Add missing examples to daily CI
Some examples are not tested by the daily CI. All examples in this repository preferably should. See https://github.com/optuna/optuna-examples/pull/1#issuecomment-837856331.
This recent contribution as well https://github.com/optuna/optuna-examples/pull/11
- [ ] kubernetes/
- [x] multi_objective/
- [x] samplers/
- [x] visualization/
- [x] enqueue_trial.py
- [x] max_trials_callback.py
- [x] quadratic_simple.py
- [ ] quickstart.ipynb
- [ ] rapids_simple.py
- [ ] ray/ray_joblib.py
- [x] simple_pruning.py
- [x] skimage/skimage_lbp_simple.py
Implemented most. Ray, Kubernetes, the two .ipynb
files, and Rapids are not implemented. Checking if Ray and the .ipynb files can be made to work. Assume testing Kubernetes within a docker or rapids not worth effort to add to CI.
Thanks for the updates, I'll have a look at #26. For the RAPIDS example, we could just CI check if it compiles. https://docs.python.org/3.8/library/py_compile.html
python -m py_compile rapids_simple.py
As for the Kubernetes example, I don't think that it's urgent, but am wondering if we shouldn't leave it as an open issue.
As suggested, made rapids a compile check only. I worked on ray_joblib.py a bit, but was unable to get the test to run on github actions. Apparently, unless running on python 3.7, it gives this error:
| ERROR: Could not find a version that satisfies the requirement ray (from versions: none)
| ERROR: No matching distribution found for ray
Even limiting to only python 3.7, I had a domain error
on in the github workflow. Google search on the error did not show a clear approach, so I'm leaving as compile only, and marking this issue contribution welcome in case someone more familiar with ray wants to work on this.
@hvy and @Crissman discussed the daily CI for Ray and RAPIDS last year, but no one investigates the kubernetes
and quickstart.ipynb
. I briefly checked them. I'm not familiar with them, so please correct me if I'm wrong.
kubernetes/
The kubernetes example uses minikube, and minikube provides the official guide to setup minikube as CI step in GitHub Actions. It may help us create daily CI for the kubernetes examples.
quickstart.ipynb
papermill
says that it enable us to execute IPython notebook via Python API as follows:
import papermill as pm
pm.execute_notebook(
'path/to/input.ipynb',
'path/to/output.ipynb',
parameters = dict(alpha=0.6, ratio=0.1)
)
It may help us create daily CI for quickstart.ipynb
. Also, we can see third-party github action to run notebook using papermil.
I'm not sure if we can use it in terms of security.
https://github.com/marketplace/actions/run-notebook
@toshihikoyanase Thank you for investigating this issue. Let me leave one comment.
quickstart.ipynb
In fact, one notebook has already been tested in this repo's CI by using jupyter nbconvert
command.
https://github.com/optuna/optuna-examples/blob/92bbe4bce97118ca6d8eab13a2a850a68b717c6a/.github/workflows/hiplot.yml#L35
@nzw0301 Thank you for your comment. Then, we can apply jupyter nbconvert
to quickstart.ipynb
!
This issue has not seen any recent activity.
This issue was closed automatically because it had not seen any recent activity. If you want to discuss it, you can reopen it freely.