Jan
Jan
I think it's a good starting point for refactoring the `Callable` potential API.
At the moment, if a user passes a just a `Callable` as potential, we test during runtime whether it has the required arguments, e.g., `theta` and `track_gradients` and `x_o` (or...
More context for resolving this: it makes sense to define `Potential` as a Python protocol, instead of an ABC. This will give more flexibility in designing custom potentials without the...
@manuelgloeckler how has this been resolved? Is it fixed in the new PyTorch release?
It would also be worth checking out https://github.com/probabilists/azula
Sorry, I was a bit too early with the issue. It's not merged yet but only fails in #1153
not relevant for package development, but branch must not be deleted!
Hi @Muhammad-Rebaal thanks for offering to take on this issue. Yes, your plan looks good. Maybe this has been raised as an issue in `pymc` already. If not, we should...
yeah, sounds good! Please check our contribution guide for how to get started contributing. Let me know if you need any input. Thanks! 🙏
Great to get this started! A couple of questions: - Where you able to reproduce the issue with the current `sbi` release, `0.25`, by running the test without the `fail`...