Thomas R. Bromley

Thanks! @mudit2812 do you have the ability to merge in `master` to the PR branch so that we can do a final round of tests?

Thanks a lot for opening this @Tarun-Kumar07! We'll give this a review soon and also help with the CI issue you mentioned. Sorry if things are a little delayed this...

Thanks @Tarun-Kumar07! Great job with this PR. We often tag contributors in our release marketing on Twitter/X (this will happen at the start of May). If you would optionally like...

Sorry for the delay in reviewing this @Qottmann, this week has been a bit busy. I'll try to get to this asap early next week, but feel free to ping...

Any worries about 7d05c63f69eb8ab364a11b2bc49027d2cefc0c30? We could always create a new PR since the changes here are pretty localized, if we're worried about a polluted branch.

Could this be fixed in a similar way to #4966?

Closing this PR as it is stale, we can return back to this later! Tracked in story 51340.

Thanks @Jaybsoni! Closing this PR as it's stale. The objective of improving performance is something we still want, but we would come back to that in a new PR.

Thanks @BorjaRequena! This will be an awesome resource for reinforcement learning :star_struck:

Will the return type of `generator` depend on whether new opmath is enabled? Or are we always returning the new type?