search

JuliaGaussianProcesses / KernelFunctions.jl

Julia package for kernel functions for machine learning
https://juliagaussianprocesses.github.io/KernelFunctions.jl/stable/
MIT License
267 stars 32 forks source link

Use dev in examples? #339

Closed willtebbutt closed 3 years ago

willtebbutt commented 3 years ago

Out examples each have a different environment, and each has it's own dependency on AbstractGPs. Currently, this dependency is not tied to the currently checked-out branch via dev ../.. or whatever. I wonder whether we should do this?

Reasons in favour:

  1. allows us to run the examples in CI to ensure that they still pass when we make changes.
  2. it ensures that our examples run on whatever the most recent version of
  3. if we do run our examples as part of CI, it'll give us a few examples of stuff running in the wild, which is extra coverage, which is probably a good thing

I also don't recall whether we've already had this conversation and ruled this out...

devmotion commented 3 years ago

The examples (or at least some of them) use the checked out development version: https://github.com/JuliaGaussianProcesses/KernelFunctions.jl/blob/fbce558b9362fa7e6d4c810ec7a98de09f3990b8/examples/kernel-ridge-regression/Manifest.toml#L284

willtebbutt commented 3 years ago

Oh -- I extrapolated from a single data point! https://github.com/JuliaGaussianProcesses/KernelFunctions.jl/blob/fbce558b9362fa7e6d4c810ec7a98de09f3990b8/examples/gaussian-process-priors/Manifest.toml#L282

I'll make a PR to correct.