search

microsoft / lightgbm-benchmark

Benchmark tools for LightGBM
MIT License
14 stars 7 forks source link

Implement continuous training by allowing an (optional) input with a previously existing model #263

Open jfomhover opened 2 years ago

jfomhover commented 2 years ago

In LightGBM training, you can use a pre-existing model. This does not exist yet in our implementation. We could introduce an optional input for continuous training.

felixnext commented 2 years ago

Likely will also be impacted by the continue train support for the transform-library: https://github.com/microsoft/lightgbm-transform/issues/18