1d and 2d cnn embedding nets.

[janfb]

failing test for 1D CNN embedding, see #750

and a new implementation of the CNN embedding net including 1D and 2D convolution, and some input checks.

[michaeldeistler]

The test looks good, thanks! But I guess we do not merge this yet, right?

[janfb]

yes, we should fix the 1D embedding net with asserts and find hyperparams that work for the common scenarios.

[michaeldeistler]

Nice, thanks for taking care of this! A few comments below.

[codecov-commenter]

Codecov Report

Merging #751 (0d66e3a) into main (fda9d4e) will increase coverage by 0.35%.
The diff coverage is 100.00%.

❗ Current head 0d66e3a differs from pull request most recent head 0f8dbf5. Consider uploading reports for the commit 0f8dbf5 to get more accurate results

@@            Coverage Diff             @@
##             main     #751      +/-   ##
==========================================
+ Coverage   74.13%   74.48%   +0.35%     
==========================================
  Files          78       77       -1     
  Lines        6081     6111      +30     
==========================================
+ Hits         4508     4552      +44     
+ Misses       1573     1559      -14     

Flag	Coverage Δ
unittests	74.48% <100.00%> (+0.35%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
sbi/neural_nets/embedding_nets.py	91.42% <100.00%> (+31.01%)	⬆️
sbi/utils/restriction_estimator.py	76.79% <0.00%> (-6.33%)	⬇️
sbi/samplers/rejection/rejection.py	87.05% <0.00%> (-2.95%)	⬇️
sbi/utils/__init__.py	100.00% <0.00%> (ø)
sbi/inference/posteriors/direct_posterior.py	98.21% <0.00%> (ø)
sbi/samplers/__init__.py
sbi/analysis/sbc.py	80.00% <0.00%> (+1.33%)	⬆️
sbi/inference/posteriors/mcmc_posterior.py	84.83% <0.00%> (+2.80%)	⬆️
sbi/inference/snre/snre_base.py	94.87% <0.00%> (+3.41%)	⬆️
sbi/inference/base.py	85.61% <0.00%> (+7.19%)	⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[michaeldeistler]

Good to go once documentation is finalized.