ramseyissa

@Balandat Thank you very much for the quick reply! Yes exactly, this is an active learning problem. My main goal is to try to minimize the amount of data (of...

Ok that makes sense. So to validate this i would just fit and predict using the GP on the validation set iteratively after selecting the next best point (as in...

@Balandat Hello again, Thank you very much for the initial advice. I have a semi-working implementation (i say semi for the following reason). I notice when i evaluate the acquisition...

Ah, that makes sense. I updated my script based on your suggestion. I went ahead and added a subset of the data mc_points = 500. So what it does is...

Hey @Balandat, thanks for clarifying what the `mc_points` are suppose to be. I sort of went down the Botorch rabbit hole these last few days and I am starting to...

EDIT: Just realized that my assumption is incorrect. It seems like the best way to go is `evaluate_acquisition_function` as it would allow for evaluating on my actual candidate list. Here...

Hey @Balandat, its been a little while, but i reevaluated my approach a bit (hopefully this is closer to what you suggested early on). See sample code please. I have...

@sgbaird, thank you very much, this was very helpful! I managed to tweak this a bit for my use case, and swapped out the qNIPV acquisition function for qMES using...

@Balandat @sgbaird : I have been trying to use qNIPV in the same way i used MES, but i keep getting an error that isn't making much sense to me....

@sgbaird, I went through the [MWE example for qNIPV](https://github.com/sparks-baird/self-driving-lab-demo/blob/main/notebooks/qnipv-mwe.ipynb) that was referenced above and it was very helpful. I believe it is the same working example you referenced in https://github.com/facebook/Ax/issues/1557#issuecomment-1528840459...