Guy Moss

Can we close this issue as resolved @janfb?

The root cause is probably not inconsistencies in ruff (will change issue name accordingly), but instead a hackathon issue of 30 people trying to learn to use ruff at the...

More context: The [SBC tutorial ](https://github.com/sbi-dev/sbi/blob/main/tutorials/13_diagnostics_simulation_based_calibration.ipynb) currently only discusses calibration of the 1D marginals of the posteriors. Simulation-based calibration can be done with any reducing function ([Talts et al.](https://arxiv.org/abs/1804.06788)). Using...

Hi all, had a look at this repo and I found the problem. the `nflows` implementation of MADE doesn't take the context into account correctly. The input to the "first"...

Thanks for setting up the issue! I will begin with 2) `likelihood_nn` and 5) integration into SNLE base classes.

> * [ ] 3) Change `classifier_nn` helper function to return ( new Ratio Estimator Class ?) Thinking about this some more, not sure how necessary this is? Currently only...

Closing this in favour of #992 as everything else is addressed already.

I've made some progress now towards this PR, and would like some feedback before I continue. > BasePotential can either "allow_iid" or not. Given `batch_dim_theta!=batch_dim_x`, we need to decide how...

> Great effort, thanks a lot for tacking this 👏 > > I do have a couple of comments and questions. Happy to discuss in person if needed. Thanks for...

Additional context: The default parameters are defined [here](https://github.com/sbi-dev/sbi/blob/main/sbi/inference/posteriors/mcmc_posterior.py#L37). Following discussion in #924 - a good suggestion for the default parameters for `MCMCPosterior` might be: `method="slice_np_vectorized", num_chains=20, warmup_steps=50, thin=5` It could...