botorch
botorch copied to clipboard
pytorch
Bayesian optimization in PyTorch
Summary: Move multitask SAAS GP into BoTorch Differential Revision: D35573439
Summary: See https://github.com/pytorch/botorch/discussions/1217 Differential Revision: D36490304
Multiple questions have popped up for this (e.g. https://github.com/pytorch/botorch/issues/1244#issuecomment-1146842945), we should really put together some docs / an example tutorial for the `InputTransform` and `OutputTransform` functionality.
In some cases we may allow not using box constraints (e.g. when optimizing Alebo). Also, in general it would be good to the optimziation raise an error if the constraint...
# 🚀 Feature Request Implement output selection in untransfrom_posterior for outcome transforms. ## Motivation Currently no outcome transforms support untransforming posteriors when outputs are selected. This makes using multi-output SingleTaskGP...
# 🐛 Bug Looks like the normalize transform is incorrectly double-batching the posterior but only after training the model. Not entirely sure where the bug is in here: ## To...
Hi, I was thinking about the possibility to use non GP models within botorch. For example to use a GP for one objective and a neural network (ensemble) for another...
Summary: Currently, we apply the input transforms in `train` mode at the `forward` call, and in `eval` model at the `posterior` call. We also use a `transform_train_inputs` call at the...
# 🚀 Feature Request Currently input transforms can be parametric and the parameters can be optimized jointly with the other hyperparameters of the GP (e.g. input warping does). Outcome transforms...
# 🚀 Feature Request If KG is applied to a heteroskedastic model, it will raise a NotImplementedError that comes from the following function. [code](https://github.com/pytorch/botorch/blob/main/botorch/models/gp_regression.py#L421) Is there any plan to implement...
