Experiments onboarding

[dberenbaum]

Report

Depends on iterative/dvclive#366.

Possible updates to reflect new experiments onboarding workflow:

• https://dvc.org/doc/dvclive/get-started#dvc-integration: should "just work" inside a dvc repo; may need to clarify what to do for an existing pipeline
• https://dvc.org/doc/start/experiment-management/experiments: update to use this simplified workflow; show steps to convert to full pipeline? What repo should we use? Do we even need a repo?
• https://dvc.org/doc/user-guide/experiment-management: update so it doesn't assume dvc exp run workflow
• dvc exp save command ref @dtrifiro (add docs for dvc exp save #4148)
• Top-level metrics/params in https://dvc.org/doc/user-guide/project-structure/dvcyaml-files (edit: and Top-level params and metrics dvcyaml-schema#26) @skshetry
• Live(save_dvc_exp=True) API ref in https://dvc.org/doc/dvclive/api-reference/live @daavoo (dvclive: Updates for "DVCLive only" experiments  #4162)
• dvclive: drop the dvclive get started. #4409

[jorgeorpinel]

I removed a couple people who seem to have contributed their parts but not sure about the remaining points.

Also is this still p1 based on those? (I guess they're kind of the most important ones 😅 -- make exp save the happy path in GS and UG).

[daavoo]

Also is this still p1 based on those? (I guess they're kind of the most important ones 😅 -- make exp save the happy path in GS and UG).

@jorgeorpinel I think It is not as much about exp save, I think It is about this workflow https://iterative.ai/blog/exp-tracking-dvc-python

[dberenbaum]

Also is this still p1 based on those? (I guess they're kind of the most important ones 😅 -- make exp save the happy path in GS and UG).

Yup, haven't gotten to this yet, but we need to overhaul the experiments guide and incorporate this workflow into the work happening on https://github.com/iterative/dvc-get-started-cv, which should replace https://github.com/iterative/example-dvc-experiments.