Jan

not sure `test_requirements.txt` is actually used, but this should fix the problem shown in #3 closes #3

https://github.com/mackelab/gatsbi/security/dependabot/1

categorial mass estimator and mnle fails for 2D density estimator test

2

**To Reproduce** run `test_correctness_of_batched_vs_seperate_sample_and_log_prob`

- [x] condense and refine change log - [x] change names to GH handles - [ ] stick with version 0.23? - [ ] do we need a short tutorial...

Improving the structure of the documentation and website

4

Our documentation has grown organically. During the March 2024 hackathon we made a lot of improvements on the content and currently we are working on improving the website accordingly, see...

fixes #60 Changes: - pass on device to `DataParallel` - add a minimal test for fid calculation with different devices - add docstrings and types for `fid.py` Note that I...

joss paper draft

1

- add paper draft - add action to test-build from markdown.

missing features and todos for score estimation

3

there are a couple of unsolved problems and enhancements for NPSE: ### MAP MAP is using the score directly for doing gradient ascent on the posterior to find to MAP....

the z-scoring tests need some refactoring: https://github.com/sbi-dev/sbi/blob/dac03dbdcda3760e3de6d843d0ea4e4b644d8dc0/tests/sbiutils_test.py#L415 - variable names are confusing, e.g., `net` should be called `build_fun` - there are many duplicate test cases - why pass `net(t_batch, t_batch)`...

In our current implementation, we do not allow the `condition` to have a `sample_dim` (only a `batch_dim`). It would be nice to have a `sample_dim` as well, e.g., for post-hoc...