-
Notifications
You must be signed in to change notification settings - Fork 394
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
DVCLive: Deprecate live
section of dvc.yaml
.
#3411
Conversation
In favor of reusing `metrics` and `plots`. Closes iterative/dvclive#229
@@ -97,13 +97,11 @@ $ dvc stage add --name train \ | |||
--params seed,lr,weight_decay \ | |||
--checkpoints model.pt \ | |||
--plots-no-cache predictions.json \ | |||
--live dvclive \ | |||
--plots-no-cache dvclive/scalars \ |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Any particular reason to use --plots-no-cache
instead of --plots
like above?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think that we should encourage plots-no-cache
in the default example usages (for dvclive/scalars
)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
No longer sure about this as default 😅 .
I was thinking it might be better to document the previous default behavior:
metrics:
- dvclive.json:
cache: True
plots:
- dvclive
And later document this flexibility in as an admonition or something.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
What do you mean "as default"? Do you mean this is an implicit recommendation? Well, up to you here @daavoo. I think -no-cache
makes more sense for small text files Git can handle but maybe that context is lost in the sample, and regular --plots
is def. easier to read.
Maybe the Q is whether
--plots
should by default not commit outputs to cache, but that's a core Q and I think it was discussed some time ago.
@@ -97,13 +97,11 @@ $ dvc stage add --name train \ | |||
--params seed,lr,weight_decay \ | |||
--checkpoints model.pt \ | |||
--plots-no-cache predictions.json \ |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Do you know if --plots-no-cache predictions.json
is used at all? Seems like that file isn't meant to be a plot and maybe should be --outs-no-cache
instead. cc @flippedcoder
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks good! I can take another look once deployments are up again.
Co-authored-by: Dave Berenbaum <[email protected]>
Hi. Why was the |
> This only has an effect when used with `--type=dl`. | ||
- `--live` - use the given path to set up `metrics` and `plots` as in |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
--type=dl
is no longer needed? Is it still a thing? That option desc. may also need an update.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The thing is that:
This only has an effect when used with
--type=dl
Was not true at all. Passing live
can have effect regardless of whether you use --type=dl
or not
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
OK. So is the --type
description accurate?
dl stages are intended for use in deep-learning scenarios, where metrics and plots are tracked with DVCLive.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It needs to be revisited now that iterative/dvc#7703 has been merged. It belongs to different P.R. though
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Commented in iterative/dvc#7703 (review)
Gatsby Cloud Build Reportdvc.org 🎉 Your build was successful! See the Deploy preview here. Build Details🕐 Build time: 1m PerformanceLighthouse report
|
This comment was marked as resolved.
This comment was marked as resolved.
Yup! Thanks @julieg18 . Cc @dberenbaum you can see it here: https://dvcorg-deprecatelive.gtsb.io/doc/dvclive |
Co-authored-by: Jorge Orpinel <[email protected]>
Co-authored-by: Restyled.io <[email protected]>
Waiting to resolve iterative/dvc#7534. |
I see. Probably doesn't have to block this PR tho? |
Yup, you are right, although it will make documenting Edit: opened iterative/dvc#7703 to move it along. |
@dberenbaum @jorgeorpinel could you take another look to see what's missing? I have addressed everything except for #3411 (comment) , which I believe requires separate P.R. to update (as |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
One sug ⬆️
Co-authored-by: Jorge Orpinel <[email protected]>
In favor of reusing
metrics
andplots
.Closes iterative/dvclive#77
Closes iterative/dvclive#229
Closes iterative/dvclive#230