Use dill to serialize logp functions in DensityDist #4053
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
aseyboldt
force-pushed
the
fix-density-dist-pickle
branch
2 times, most recently
from
0da0c00
to
2ec28db
Compare
aseyboldt
force-pushed
the
fix-density-dist-pickle
branch
from
2ec28db
to
a92ce54
Compare
Codecov Report
@@ Coverage Diff @@
## master #4053 +/- ##
==========================================
+ Coverage 86.82% 86.83% +0.01%
==========================================
Files 88 88
Lines 14154 14165 +11
==========================================
+ Hits 12289 12300 +11
Misses 1865 1865
|
twiecki
reviewed
Aug 16, 2020
RELEASE-NOTES.md
Outdated
| @@ -5,6 +5,7 @@ | |||
| ### Maintenance | |||
| - Mentioned the way to do any random walk with `theano.tensor.cumsum()` in `GaussianRandomWalk` docstrings (see [#4048](https://github.com/pymc-devs/pymc3/pull/4048)). | |||
| - Fixed numerical instability in ExGaussian's logp by preventing `logpow` from returning `-inf` (see [#4050](https://github.com/pymc-devs/pymc3/pull/4050)). | |||
| - Use dill to serialize user defined logp functions in `DensityDist`. `dill` is now a required dependency. (see [#3844](https://github.com/pymc-devs/pymc3/issues/3844)). | |||
There was a problem hiding this comment.
Suggested change
| - Use dill to serialize user defined logp functions in `DensityDist`. `dill` is now a required dependency. (see [#3844](https://github.com/pymc-devs/pymc3/issues/3844)). | |
| - Use dill to serialize user-defined logp functions in `DensityDist`. `dill` is now a required dependency. (see [#3844](https://github.com/pymc-devs/pymc3/issues/3844)). |
There was a problem hiding this comment.
I would add what this allows what wasn't possible before. did pickling just throw an error?
twiecki
reviewed
Aug 16, 2020
scripts/create_testenv.sh
Outdated
| @@ -24,13 +24,15 @@ ENVNAME="${ENVNAME:-testenv}" # if no ENVNAME is specified, use testenv | |||
| if [ -z ${GLOBAL} ]; then | |||
| if conda env list | grep -q ${ENVNAME}; then | |||
| echo "Environment ${ENVNAME} already exists, keeping up to date" | |||
| source activate ${ENVNAME} | |||
There was a problem hiding this comment.
I think new style is conda activate (or mamba).
There was a problem hiding this comment.
If it's brittle probably best to keep it the way you had then.
There was a problem hiding this comment.
Well, it is now the way the conda docs specify it...
But I can of course change it back.
aseyboldt
force-pushed
the
fix-density-dist-pickle
branch
from
f9b563d
to
712e3fb
Compare
twiecki
approved these changes
Aug 16, 2020
|
Thanks! |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fix #3844
When users pass functions to
pm.DensityDist, then in most cases pickle will not be able to deal with those functions. The code will still work on Linux, because the default multiprocessing context isfork, and we don't even have to pickle the logp function. On Windows and Mac the default isspawnorforkserverhowever, and code usingpm.DensityDistwith a logp function that is not accessible in a module or protected byif __name__ == '__main__will fail. This includes all functions defined in a notebook as far as I understand.We can work around this issue by manually serializing the logp function with dill.
This PR introduces dill as a new required dependency to PyMC3, before it was optional.
Edit: I had some trouble with the cache of miniconda, so I also included a fix so that the cached environment is updated.