Open OriolAbril opened 3 years ago
@ckrapu, feel free to decline. We are currently trying to update all the notebooks to make sure they follow pymc3 best practices and then have them run on pymc3 4.0+. We are tracking our progress in this project, it has a detailed description here. We submitted a project proposal to outreachy to have some extra help with that and therefore have a smooth transition from pymc3 v3 to pymc3 v4 with great docs on v4.
No problem, I'm cleaning it up as we speak. This was a good chance for me to clean up the priors in the notebook so that there wouldn't be any need to filter out bad samples from the posterior for the gaussian process. As an aside, do you know how to suppress this warning? It's showing up 10-15 times per usage of sample
.
/Users/v7k/anaconda3/envs/pymc3_dev/lib/python3.8/site-packages/aesara/graph/fg.py:500: UserWarning: Variable Elemwise{pow,no_inplace}.0 cannot be replaced; it isn't in the FunctionGraph
It looks like you are using aesara already which may be a bit too early (as pymc3 backend), which pymc3 versions are you using?
I'd been using 3.11.1. Should I run a version that's backed by Theano?
3.11.1 should run on theano-pymc, not yet on aesara. Here is the link to it's requirements file: https://github.com/pymc-devs/pymc3/blob/v3.11.1/requirements.txt
I must have gotten my environments mixed up, then. Let me try with a clean conda env.
File: https://nbviewer.jupyter.org/github/pymc-devs/pymc-examples/blob/main/examples/case_studies/log-gaussian-cox-process.ipynb Reviewers: @ckrapu
Context
Known changes needed
Changes listed in this section should all be done at some point in order to get this notebook to a "Best Practices" state. However, these are probably not enough! Make sure to thoroughly review the notebook and search for other updates.
General updates
Changes for discussion
Changes listed in this section are up for discussion, these are ideas on how to improve the notebook but may not have a clear implementation, or fix some know issue only partially.
ArviZ related
from_pymc3_predictions
to filter nans and slice/reduceintensity_samples
Notes
Exotic dependencies
None
Computing requirements
Model takes roughly 5 mins to sample.