ludwig

ludwig copied to clipboard

Added callbacks for k-fold cross validation

1

Also adds `cross_validation` section to the schema. Doesn't have much affect for now, but in the future we may want to expand the role of cross validation, so putting it...

tgaddair

Schemify initializers

1

tgaddair

Bump py39 test to Ray 2.2

2

geoffreyangus

Allow early stop based on walltime since improvement

1

This PR: - stores the Posix timestamp of the last improvement in ProgressTracker - adds a convenience callback implementing `should_early_stop` that returns True if the given timeout has been reached...

jppgks

tgaddair

WIP: Pytorch 2.0 compiled model

1

tgaddair

[fix] Only drop rows where Nan is not in target column

1

This PR fixes the following bugs in preprocessing. Both bugs exist because we do joint preprocessing of `training_set`, `val_set` and `test_set` by concatenating then and passing them to `build_dataset`. -...

ShreyaR

This PR begins the work required to add S4 ([paper](https://arxiv.org/abs/2111.00396), [code](https://github.com/HazyResearch/state-spaces/blob/main/src/models/sequence/ss/standalone/s4.py)) to Ludwig. The implementation work thus far has focused on adapting the high-level API of the `S4` class to...

geoffreyangus

tgaddair

This PR contains the following (so far): 1. Enable loading of time series data from column-major as well as row-major sources (specified by a config option). If column-major, all input...

nimz