Carl Goodwin

Illustrating how the relative cost grows with an increasing number of columns: ``` r library(tidyverse) library(microbenchmark) data % group_by(group) results % filter(all(!is.na(value1))), none1 = data %>% filter(none(value1, is.na)), all2 =...

This feature would be very helpful.

Hi @cregouby - I've tried progressively reducing `virtual_batch_size` to 16, but `tabnet_fit` still hangs. `tabnet_pretrain` runs using the defaults. Both produce the message. I'm using an M2 Max with 64...

I get this error too and have been able to reproduce it with a toy example. If the `step_dummy()` line is uncommented, then it works. lightgbm does though support categorical...

Hi Simon - Would it be possible for stacks to ignore the 0-member solutions if there are other valid solutions? With my real data some mixture / penalty combinations had...

Hi @cregouby - Thank you. Ahead of trying it on my own data, I've made a quick test using the toy `lending_club` data. Untuned TabNet and XGBoost models, with and...

In 0.6.0.9000 I'm getting the message `Configured `weights` will not be used`: (It's the same example as per above where the weights were being passed along from the workflow.) ```...

Hi @cregouby - Thank you for clarifying. If it's possible to set case_weights more than one way, e.g. in a tidymodels `workflow()` and also in `tabnet_fit()`, then maybe the message...

I was just about to post the same issue. Adding reprex here in case it helps: ``` r library(tidyverse) library(scales) # Error tibble(x = seq(1, 10, 1), y = seq(10,...