Ryan Friedman

Early stopping and other parameters for learning rate scheduler

1

#### Reference Issues/PRs Fixes #183 #### What does this implement/fix? Explain your changes. 1. Added support for other parameters for `ReduceLROnPlateau`. The `TrainModel` object has a new argument `scheduler_kwargs` that...

Support for other schedulers and early stopping

2

I was hoping to use a different learning rate scheduler than the default implementation (reduce by 0.8 after a validation loss plateau with patience of 16 epochs). Unfortunately it seems...

UMAP fails with later version of irlba package

1

I upgraded from R/4.3.1 to R/4.3.2 and also upgraded my packages. When I did, I got the following error from UMAP: `Error in irlba::irlba function 'as_cholmod_sparse' not provided by package...

I've found that in the case of `train_model`, none of logging handlers ever close (i.e. `selene_sdk.train_model.log`, `selene_sdk.train_model.train.txt`, and `selene_sdk.train_model.validation.txt`). I assume this happens in `evaluate_model` too but I haven't looked...

The develop branch has been updated to source a `$HOME/.monoclerc` file on load. However, when I call `convert_counts_matrix(cds)` with no additional arguments, I get the following error because the `matrix_control`...

Hi, I've noticed that I can get inconsistent results on identical runs of PLN on the same data, even when setting a seed in R. There appear to be two...

When a CDS object is saved with `save_monocle_objects()`, the tarball contains the full absolute path. This is an issue because when you untar the object, the actual CDS can be...

It would be convenient to use `skrrrahh()` to wrap a function call, get the audio notification, and then return the result of the function call. Currently, the `skrrrahh()` returns an...