Release 0.15.0

[BenjaminBossan]

This is going to be a rather small release. However, since I'll do a presentation about skorch in 2 weeks at PyData Amstersdam, it would be nice to have the bugfixes related to accelerate, as this will be part of the talk.

• Update VERSION
• Remove code for backwards compatibility with v0.11 (see Loading extra arguments w/ cuda dependency on CPU #880): The plan was to remove this 2023-09. We're almost there, I think it's fine to remove it a bit earlier.
• Update CHANGES: Because of the above, I also added an entry on how to transition these pickle files.

---

Proposed release text:

This is a smaller release, but it still contains changes which will be interesting to some of you.

We added the possibility to store weights using safetensors. This can have several advantages, listed here. When calling net.save_params and net.load_params, just pass use_safetensors=True to use safetensors instead of pickle.

Moreover, there is a new argument on NeuralNet: You can now pass use_caching=False or True to disable or enable caching for all callbacks at once. This is useful if you have a lot of scoring callbacks and don't want to toggle caching on each individually.

Finally, we fixed a few issues related to using skorch with accelerate.

Thanks Zach Mueller (@ muellerzr) for his first contribution to skorch.

Find the full list of changes here: v0.14.0...v0.15.0

[thomasjpfan]

The fixes and release makes sense to me. LGTM

[BenjaminBossan]

Thanks Thomas. I updated the date before merging.