Persist job artifacts to S3

[coder46]

Persistent storage of job runs, their logs, and output after job completion to a configurable S3 destination. Create helper functions in the hydra library to allow the user to easily import them into their experiment codes

Note : Also look at mlfow artifacts if it can be used https://mlflow.org/docs/latest/tracking.html#id63

[igorgad]

Hi, thanks for this great project. I was looking for something like this for a while. Just to complement this request, it would be great to also have this feature on the GCP. Even greater if we could use NFS shares instead of downloading the dataset for each experiment, but from my perspective, it's impossible to use NFS shares on ai-platform custom containers, only on GKE clusters.

Let me know if something like this is possible, then I'll create a feature request. Regards,.

[coder46]

Hi @igorgad ! Thanks for your interest and finding our platform useful for you 😊 - and definitely we will also implement this feature for GCP too 👍 . Feel free to submit a feature request. Regarding NFS shares, i think that's a really smart idea to speed up job execution as it'll save time spent on downloading large artifacts ! Also lemme see if we can do this for AWS Batch