Miles Olson
Miles Olson
Summary: As requested by David to improve reliablity on aggregation metrics Differential Revision: D36877430
Summary: Add a context manager to speed up gpytorch via their provided gpt_settings object. We utilize 3 of their contexts: fast_computations, fast_pred_samples, and fast_pred_var which you can read more about...
Summary: In this diff we create an interactive loop API (based loosely off the existing Loop API) in order to facilitate candidate pregeneration. This can be useful in many contexts,...
Summary: Something something LinearOperator Differential Revision: D39546089
Summary: Move the interactive loop away from requiring AxClient. Use new generic infra to replicate behavior of previous diff in a new method named `interactive_optimize_with_client`. Differential Revision: D39545664
Summary: Spaces in names break parsing needed for contraints in AxClient. We discussed many options fo rmoving forward here but this simple solution is the least disruptive Differential Revision: D56576451
Summary: As titled. This solves a relatively common problem (which may crop up again in some recently planning changes) in which the init args of some serializable Ax object is...
Summary: Apply is_valid_name to all parameter and metric names Differential Revision: D56316082
Summary: is_valid_name will be used for Metric and Paramter name validation. Constricting the values allowed will help with both parsing (as used in AxClient's constraints), plotting/report building, and more. Differential...