Fix handling of prior terms in ExactMarginalLogLikelihood #2039
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Outdated suggestions cannot be applied.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.
This fixes an issue originally discovered in pytorch/botorch#1259. The bug relates to handling of prior terms in https://github.com/cornellius-gp/gpytorch/blob/master/gpytorch/mlls/exact_marginal_log_likelihood.py#L42-L43
When training batched model of
n_batch
batches, theres
term (which is the log probability of observations under the prior) has shapen_batch
. The prior termprior.log_prob(closure(module))
also has a shape that isn_batch x (potential other dimensions)
. When we add the sum of all prior terms viares.add_
we end up adding each prior term to each one of the terms inres
, when they should only be added to the term corresponding to the same batch. This is corrected by replacing thesum()
with a series of sums that reduce the prior term to have the samendim
asres
.Test plan:
Units
cc @Balandat