Closed EthanMarx closed 1 month ago
@wbenoit26 let me know what you think - was able to get the cluster deployed but can't testing the actual tune jobs until the train container gets rebuilt and pushed with the updated environment
Yeah there's probably a bug or two, but shouldn't affect the rest of the tasks
I stripped out our tuning scripts and generalized things a bit into
lightray
which is also now on pypi, the motivation being to be able to incorporate it into AMPLFI as well.This PR removes our tuning modules in favor of using
lightray
This PR also
Path.home() / .ssh / id_rsa
charts
directory and github workflow which was moved tolightray
- now we use the remote chart published on that repos github for deploying the ray cluster