-
Notifications
You must be signed in to change notification settings - Fork 392
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
fix: add metric example to xgboost docs #4917
Conversation
Thanks @djsauble! As long as we are making changes, can we fix it up a bit more? For example, can it follow a workflow that looks like the xgboost get started docs? It seems like that's what they see as their primary interface, and subjectively it looks easier to understand to me. It would also be nice to flesh out the top example even more to show where other variables like |
@dberenbaum Just to clarify, you're saying we should use something like this:
…instead of…
?
This is the working XGBoost code I ended up with after going through our existing docs + the hint about metrics from @mattseddon:
How much of this do you want to see in our docs? I don't really think we should go all the way back to a specific dataset. The focus here should be on the call to |
Fair points, @djsauble. I don't want to block the improvements you made, so I just left one minor comment to clarify and otherwise it looks good, thanks! |
@dberenbaum I actually agree with you that using I'll fix this up a bit more and ping you again for a quick review. |
If we only show that |
Let's choose one or the other. The sklearn interface is probably more familiar to people, but I don't feel strongly about this. It looks like both interfaces support callbacks, which is the only hard requirement for DVCLive. Which interface do you want to go with? |
Sounds good. Let's go with the sklearn interface then. |
@dberenbaum Switched to the sklearn interface and tested all code snippets locally. Think this is good now. |
Co-authored-by: Restyled.io <[email protected]>
Fixes: iterative/dvclive#721
The XGBoost docs have a dangling
param
variable that isn't defined anywhere. Since this is where you define your metric, and this is an important part of experimentation, I added an example of how to use it to set a metric via theeval_metric
param.