Hello, I’m bringing up the topic from our earlier discussion regarding dependencies.

Currently, the project uses rather strict dependency version constraints. Can we loosen these requirements and allow a wider range of compatible versions?

Current constraints:

requires-python = ">=3.10,<3.13"

dependencies = [
"numpy~=1.26.4",
"pandas~=2.2.2",
"river",
"scikit-learn~=1.5.0",
"sortedcontainers>=2.4.0",
"torch==2.2.2",
"tqdm~=4.66.5",
"torchviz~=0.0.2",
]

Which version ranges should we target, and how should we test them?