Max Balandat

Add batch_shape property to GP model class

1

Implements #2301.

These are failing locally for me - may need to be tied to the linop version.

[Feature Request] Add batch_shape property to GP models

2

# 🚀 Feature Request Expose a `batch_shape` property on models. For Exact GPs or models that have all the training inputs, this can just return the batch shape of the...

[Bug] `batch_cross_validation` should not always pass `train_Yvar` to the model

1

# 🐛 Bug [`batch_cross_validation`](https://github.com/pytorch/botorch/blob/89f923d9f677cb7664909b2ff0bcc30b4d37007a/botorch/cross_validation.py#L151) passes `train_Yvar=None` into classes that don't use the (e.g. SingleTaskGP). Instead, it should only pass in `train_Yvar` if it is not `None`. ## Additional context Originally...

Avoid writing a "|" file in `run_tutorials` when runnnig a tutorial

3

I don't think this is necessary, and it creates a "|" file upon running tutorials that's annoying to deal with in git.

[Feature Request] Random restart optimization for MAP estimation

4

# 🚀 Feature Request Make it easier to perform random restart optimization for MAP estimation. Currently this isn't really exposed and would require some amount of custom code to set...

[Feature Request] Add input constructors for all acquisition functions

1

# 🚀 Feature Request Some of the recently added acquisition functions (e.g. from #1458) do not have input constructors for Ax. Let's add them to make sure they can easily...

`fit_gpytorch_model` has been deprecated and superseded by ``fit_gpytorch_mll`. Let's remove it for the 0.9.0 release.

[Feature Request] Make PosteriorTransform API (optionally) take features X

4

# 🚀 Feature Request Allow the `PosteriorTransform.evaluate()` and `PosteriorTransform.forward()` (https://github.com/pytorch/botorch/blob/242ddd9efc158c1e00c0d0296151891f8922dbc7/botorch/acquisition/objective.py#L48-L70) to also accept the features `X` in addition to the values `Y` or posterior, respectively. ## Motivation See #1545 ##...

Support parsing tensors without requiring tensorboard

3

I find it a bit odd that tbparse requires tensorboard in order to parse `tensor` type logs, even if those are generated e.g. by pytorch. Ideally this could be supported...