https://github.com/GPflow/GPflow
Revision b4e6298c295c6ca90f04ecd69cceeccf450b59ea authored by st-- on 27 April 2021, 16:17:09 UTC, committed by GitHub on 27 April 2021, 16:17:09 UTC
GPflow had already been optimised for speed in the training loop. This PR addresses the speed of predictions and resolves (partially) #1599. We achieve this by introducing new Posterior objects that can cache the results of a "precompute" stage (everything that's independent of test points) and thereby significantly speeds up repeated predictions. Co-authored-by: ST John <st--@users.noreply.github.com> Co-authored-by: stefanos <stefanosele@users.no-reply.github.com> Co-authored-by: John Mcleod <43960404+johnamcleod@users.noreply.github.com>
1 parent a8ee4ae
Tip revision: b4e6298c295c6ca90f04ecd69cceeccf450b59ea authored by st-- on 27 April 2021, 16:17:09 UTC
Posterior object for fast predictions (#1636)
Posterior object for fast predictions (#1636)
Tip revision: b4e6298
File | Mode | Size |
---|---|---|
conditionals | ||
config | ||
covariances | ||
expectations | ||
inducing_variables | ||
kernels | ||
likelihoods | ||
models | ||
monitor | ||
optimizers | ||
quadrature | ||
utilities | ||
__init__.py | -rw-r--r-- | 1.1 KB |
base.py | -rw-r--r-- | 8.9 KB |
ci_utils.py | -rw-r--r-- | 2.0 KB |
kullback_leiblers.py | -rw-r--r-- | 5.8 KB |
logdensities.py | -rw-r--r-- | 3.3 KB |
mean_functions.py | -rw-r--r-- | 6.0 KB |
posteriors.py | -rw-r--r-- | 23.3 KB |
probability_distributions.py | -rw-r--r-- | 1.8 KB |
py.typed | -rw-r--r-- | 27 bytes |
types.py | -rw-r--r-- | 703 bytes |
versions.py | -rw-r--r-- | 177 bytes |
Computing file changes ...