Grid search with ray tune #50
Open
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
cdancette
force-pushed
the
grid-search
branch
2 times, most recently
from
f84d227 to
d7071f7
Compare
Grid search is called like this: python -m bootstrap.grid -o <path> --gpu 0.5 --cpu 10 This will run training on all available gpus and cpus, with 10 cpus per training and 0.5 gpus assigned for each training (2 jobs per gpu). The option file must contain a new option: "gridsearch:" which contains a list of options to modify, and their associated values should be a list containing all the values to test. See the example file in tests/options-grid.yaml TODO: analysis at the end. This will need an api to get the best results for a given run, which exists only as a script for now (compare.py)
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Ray tune makes it possible to run grid searches with multiple gpus, and multiple jobs per gpus.
Grid search is called like this:
This will run training on all available gpus and cpus, with 10 cpus per training and 0.5 gpus assigned for each training (2 jobs per gpu).
The option file must contain a new option: "gridsearch:" which contains a list of options to modify, and their associated values should be a list containing all the values to test.
See the example file in
tests/options-grid.yamlNote that there is no analysis at the end. This will need an api to get the best results for a given experiment. which exists only as a script for now (compare.py).
But this is not essential given that we already have the compare that does this for us
When this is ready to go, I'll add some documentation