Grid search #386
Grid search #386
Conversation
|
I see how the error is happening -- the ci/cd server seems to be trying to compile all .py files, including the one you are using as a template for the encoder. Perhaps changing the suffix from .py to something like .template would work. |
|
I am also shocked that 54 files have changed -- but most of these seem to be whitespace changes. It would be better for me as a reviewer if such reformatting occurred in a separate pull request. |
tests/test_grid_sampler.py
Outdated
| params["x2"] = {"type": "boolean", "default": True} | ||
|
|
||
| # python file is its own template | ||
| encoder = uq.encoders.GenericEncoder('tests/grid_search/test_grid.py', |
There was a problem hiding this comment.
perhaps change the name to 'tests/grid_search/test_grid.template' so that the invalid program 'tests/grid_search/test_grid.py' is not checked by the CI/CD system
There was a problem hiding this comment.
Right, I didn't know it would try to complile these. Will change to .template indeed
There was a problem hiding this comment.
And sorry for the 54 files, shouldn't have done the pep8 fixes
tests/grid_search/test_grid.py
Outdated
| @@ -0,0 +1,14 @@ | |||
| #!/usr/bin/python3 | |||
There was a problem hiding this comment.
I suggest renaming this file to 'tests/grid_search/test_grid.template' so that the CI/CD tests don't fail when trying to see if the syntax is OK.
| try: | ||
| fabsim("fetch_results", "", machine) | ||
| return True | ||
| except: |
Check notice
Code scanning / CodeQL
Except block handles 'BaseException'
There was a problem hiding this comment.
Yes, you should never use try..except like this (with no exception type specified)
| #get the output lines of fab <machine> stat | ||
| try: | ||
| out = subprocess.run(['fabsim', machine, 'stat'], stdout=subprocess.PIPE) | ||
| except: |
Check notice
Code scanning / CodeQL
Except block handles 'BaseException'
Added a
Grid_samplerto perform hyperparameter tuning for neural networks in parallel. New files:easyvvuq/sampling/grid_sampler.py: the new sampler classtests/test_grid_sampler.py: a test for the grid samplertutorials/hyperparameter_tuning_tutorial.ipynb: a tutorial on using the GridSampler to tune the hyperparameters of a TensorFlow neural network for character recognition.tutorials/hyperparameter_tuning_tutorial_with_fabsim.ipynb: same as above, only using FabSim3 to submit the ensemble.I tested it locally and on Archer2, using FabSim + QCG PilotJob.