Vincent N
unread,Jul 6, 2021, 3:20:04 AM7/6/21Sign in to reply to author
Sign in to forward
You do not have permission to delete messages in this group
Either email addresses are anonymous for this group or you need the view member email addresses permission to view the original message
to sig-tfx-addons, Michael Hu, rober...@google.com, Vincent N, sig-tfx-addons, Daniel Kim, Gerard Casas Saez
> Is the current direction for the XGBoost project to include a dependency on BQML?
I don't think so. Although if the evaluator is built based on xgboost native model loader instead of pickled sklearn Pipeline, it could work with exported BQML XGBoost models.
For the local pipeline here are 2 example changes that follow different approach, let me know which one you prefer:
2.
https://github.com/cent5/tfx-addons/commit/4f95a51d067ed0274745082459ed5b96c7b74e86 - This is using xgboost native API, which would require a different evaluator (so I think it might makes more sense to have this pipeline in a separate dir altogether). Instead of pickling the sklearn Pipeline object, we use xgboost support functions to save and load the model. This serialization format is also compatible with CAIP, plus other XGBoost interfaces such as JVM, C++, etc. So it provides more flexibility when building serving infra.