Sait Cakmak

> Is there a built-in way in Ax’s modular BoTorch backend to jointly model multiple outputs? Technically, yes -- though I haven't tried this in quite a while. If you...

Hi @leolin8806 & @leonardoguilhoto. I spent some time looking into this. There isn't a nice off-the-shelf way of doing this, but with some customization, I got something working. Note that...

I think this is great! What I had done was to take out linear operator and any parts of GPyTorch that we didn't need for ExactGPs and create a bare-bone...

The memory issue might be related to https://github.com/pytorch/botorch/issues/2310. Due to some odd broadcasting within PyTorch / GPyTorch, the ensemble models can consume excessive amounts of memory when evaluated with large...

The tensor shapes in that operation, particularly the shape of `x2` is pretty large. Do you know why the -2 / q-batch dimension is 1611? I am guessing it is...

> 32 × 50 = 1600 fantasy points I am confused about why these 1600 points end up in the q-batch (-2) dimension. When we look at the tensor shape...

After writing this, I noticed that I am looking at the latest code rather than the stable 0.14.0. `@average_over_ensemble_models` used to be part of `@t_batch_mode_transform`, now it is a separate...

Yes, it should be ok to remove the t-batch decorator

Yes, that'd be the correct input batch shape. You're adding additional batch dimensions (due to fantasies) to an already batched model (fully bayesian), so it is natural to run into...

Hi @suttergustavo. The error goes away if you specify `validate_task_values=False` when constructing the model. However, the behavior might be slightly different from what you had prior to 0.16.0 (https://github.com/meta-pytorch/botorch/pull/2960 specifically)....