Jan
Jan
At the moment it seems we do not have a systematic suite of tests for the growing number of sampling methods. Or, they are scattered in different test modules, e.g.,...
As an additional calibration test, we would like to implement local tests from [Zhao et al.](https://proceedings.mlr.press/v161/zhao21b/zhao21b.pdf). For this, we already have an old draft that never got merged: #379 Subtasks:...
See #413
This goes back to #51. #51 was closed because it relates only to where to reset to when retraining from scratch. This issue is about providing more choices for the...
With SNPE we can use multidimensional simulator outputs `x`. When we then use `mcmc_method={"slice", "hmc", "nuts"}` this will be stuck in an endless loop during warmup. Probably, because pyro does...
## Description: As part of our ongoing effort to ensure that our documentation is both user-friendly and current, we're seeking contributors, especially those new to SBI and our package, to...
It would be great to have FMPE implemented in `sbi`. There are implementations available that we could depend on, e.g., - dingo: https://github.com/dingo-gw/dingo/blob/FMPE/dingo/core/posterior_models/flow_matching.py - lampe via zuko: https://lampe.readthedocs.io/en/stable/tutorials/fmpe.html To Do:...
**Describe the bug** `base_posterior` raises an `AssertionError` when it is passed a `potential_fn` that does not have `x_o` as an argument, e.g., when using the `ConditionalPotential` https://github.com/sbi-dev/sbi/blob/17f3033aa55514ac6c039da729ee84ce32eb20d9/sbi/inference/posteriors/base_posterior.py#L47-L54 **To Reproduce** `pytest...
The `SNPE_A.build_posterior(...)` method overrides that of the base class to correct for the proposal posterior. **Problem:** As a side effect, it has a different signature than the base class method,...
**Is your feature request related to a problem? Please describe.** Given the new `DensityEstimator` setup (#952 , #965) we can now start implementing new density estimators. **Describe the solution you'd...