Dask is a flexible library for parallel computing in Python. The documentation will
split between the two areas of writing distributed training using the
Dask framework and
creating both the container and
yaml spec to run the distributed workload.
This is a good choice when you want to use
LightGBM or have
data parallel tasks for very large datasets where the data can be partitioned.