Fix LocScaleReparam, log params in backtest() #2365
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
| if isinstance(result[name], (int, float)): | ||
| logger.debug("{} = {}".format(name, result[name])) | ||
| for name, value in pyro.get_param_store().items(): | ||
| if value.numel() == 1: |
There was a problem hiding this comment.
make the numel boundary toggleable so that users have an easy way to inspect params? e.g. they can even choose inf
There was a problem hiding this comment.
I'm trying to avoid logging any Tensors, instead logging only Python objects. One reason for this design choice is that a single big tensor renders the entire data structure unreadable (visually, at a terminal). One option that would still satisfy this design constraint is to log scalar statistics of each non-scalar tensor, e.g. min, max, median, mean. That seems like a reasonable feature for a follow-up PR.
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Outdated suggestions cannot be applied.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.
Fixes a naming bug in
LocScaleReparamwhereby all loc-scale-reparameterized sites shared a single centeredness parameter. After this PR each such site learns a separate centeredness parameter (as originally intended).Also adds param logging to
contrib.forecast.evaluate.backtest(), which logging led me to discover the loc-scale bug. After this PR all scalar params will be printed and saved (as floats) to backtest results. My immediate motivation is to inspect global parametersstability,skewanddfofStableandStudentTdistributions.Tested