Skip to content

Commit 7e9fb1b

Browse files
saitcakmakfacebook-github-bot
authored andcommitted
Changelog for 0.8.2 (#1700)
Summary: Pull Request resolved: #1700 -- Reviewed By: esantorella Differential Revision: D43546959 fbshipit-source-id: e062f767d9fb84e19ba35d213b37767f1f6bbfb1
1 parent 1c0747c commit 7e9fb1b

File tree

1 file changed

+42
-0
lines changed

1 file changed

+42
-0
lines changed

CHANGELOG.md

Lines changed: 42 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -2,6 +2,48 @@
22

33
The release log for BoTorch.
44

5+
## [0.8.2] - Feb 23, 2023
6+
7+
#### Compatibility
8+
* Require PyTorch >= 1.12 (#1699).
9+
10+
#### New Features
11+
* Introduce pathwise sampling API for efficiently sampling functions from (approximate) GP priors and posteriors (#1463).
12+
* Add `OneHotToNumeric` input transform (#1517).
13+
* Add `get_rounding_input_transform` utility for constructing rounding input transforms (#1531).
14+
* Introduce `EnsemblePosterior` (#1636).
15+
* Inducing Point Allocators for Sparse GPs (#1652).
16+
* Pass `gen_candidates` callable in `optimize_acqf` (#1655).
17+
* Adding `logmeanexp` and `logdiffexp` numerical utilities (#1657).
18+
19+
#### Other changes
20+
* Warn if inoperable keyword arguments are passed to optimizers (#1421).
21+
* Add `BotorchTestCase.assertAllClose` (#1618).
22+
* Add `sample_shape` property to `ListSampler` (#1624).
23+
* Do not filter out `BoTorchWarning`s by default (#1630).
24+
* Introduce a `DeterministicSampler` (#1641).
25+
* Warn when optimizer kwargs are being ignored in BoTorch optim utils `_filter_kwargs` (#1645).
26+
* Don't use `functools.lru_cache` on methods (#1650).
27+
* More informative error when someone adds a module without updating the corresponding rst file (#1653).
28+
* Make indices a buffer in `AffineInputTransform` (#1656).
29+
* Clean up `optimize_acqf` and `_make_linear_constraints` (#1660, #1676).
30+
* Support NaN `max_reference_point` in `infer_reference_point` (#1671).
31+
* Use `_fast_solves` in `HOGP.posterior` (#1682).
32+
* Approximate qPI using `MVNXPB` (#1684).
33+
* Improve filtering for `cache_root` in `CachedCholeskyMCAcquisitionFunction` (#1688).
34+
* Add option to disable retrying on optimization warning (#1696).
35+
36+
#### Bug fixes
37+
* Fix normalization in Chebyshev scalarization (#1616).
38+
* Fix `TransformedPosterior` missing batch shape error in `_update_base_samples` (#1625).
39+
* Detach `coefficient` and `offset` in `AffineTransform` in eval mode (#1642).
40+
* Fix pickle error in `TorchPosterior` (#1644).
41+
* Fix shape error in `optimize_acqf_cyclic` (#1648).
42+
* Fixed bug where `optimize_acqf` didn't work with different batch sizes (#1668).
43+
* Fix EUBO optimization error when two Xs are identical (#1670).
44+
* Bug fix: `_filter_kwargs` was erroring when provided a function without a `__name__` attribute (#1678).
45+
46+
547
## [0.8.1] - Jan 5, 2023
648

749
### Highlights

0 commit comments

Comments
 (0)