Machine learning

Current implementation of Go binding can not specify options.

GPUOptions struct is in internal package. And go generate doesn't work for protobuf directory. So we can't specify GPUOptions for NewSession.

🚀 Feature

Support to pickle a jitted function (or at least throw a TypeError when using protocol 0 and 1).

Motivation

Trying to pickle a jitted function either raises TypeError: cannot pickle 'torch._C.ScriptFunction' object when protocol>1 or far worse when using protocol=0 or protocol=1 python 3.9.5 dies with:

terminate called after throwing an instance of 'std::runt

Describe the issue linked to the documentation

The description of cross-validation estimator does not explain whether the final model parameters are estimated on the entire training set, using the optimal hyperparameter obtained through cross-validation.

There's a post going around on Twitter for how to write a super fast DFA, e.g. for UTF8 validation: https://twitter.com/_rsc/status/1413843059972923394?s=19. We have a UTF8 validation function, so we should try this technique. Should be reasonably simple to do.

Currently many more Python projects like dask and optuna are using Python type hints. With the Python package of xgboost gaining more and more features, we should also adopt mypy as a safe guard against some type errors and for better code documentation.