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.

🚀 The feature, motivation and pitch

JIT serialization & freezing uses equality checks for things like common subexpression elimination (CSE).

Since aten::equal isn't implemented for sparse tensors, sparseTensorA.equals(sparseTensorB) fails, preventing CSE during freezing.

repro:

import torch

x = torch.rand((3, 4)).to_sparse()
y = torch.rand((3, 4))
print(torch.equal(x, 

Describe the issue linked to the documentation

Link the site https://scikit-learn.org/scikit-learn-benchmarks/ and https://github.com/scikit-learn/scikit-learn-benchmarks in the homepage https://scikit-learn.org.

Follow-up on #17026

It might be useful to have rationalize(x) return x if x is already rational, to allow generic code that takes both reals and rationals as inputs.

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.