Matt Warkentin
Matt Warkentin
No problem! This has been a good learning experience for me. I've enjoyed the process of contributing to this package, so I am interested in helping out with this issue...
Hmm, I don't have another motivating example off the top of my head. Thinking about how to implement SGL was the first time I realized we would need a column-to-column...
@hfrick Is this feature still in scope? Same as for #131.
Absolutely! Happy to take a look. Will report back.
I don't think I have a strong opinion on the nesting/non-nesting for single row per-observation predictions. I do agree with @hfrick that format consistency is a nice feature. I think...
I came here to ask the very same thing. It *feels* like this should work, but it doesn't. Would love for the addition of `ggplot`-friendly scales. ``` r library(tidyverse) library(clock)...
Related to the consistency of arguments, should the newly-added `repeats` argument for `group_vfold_cv()` come after `v` in the arg list (i.e., before `balance`), as is the case for its non-grouped...
Are you looking to create native `torch` implementations of any metric added to `luz`? I wonder if the `yardstick` package could possibly be used for adding immediate support for a...
Sorry, just so I understand, does **streaming mode** mean that batch-wise calculations are `update()`d along the way, but the final metric calculation is done at the end of the epoch?...