I'm working on a neural proxy function for HMCECS and have a Taylor expansion proxy with an approximate Hessian. However, the file is becoming somewhat unruly as the proxies are currently in hmc_gibbs.py. If I move (only) the proxy functions to a separate file under contrib and keep using the static method interface for HMCECS (i.e., HMCECS.taylor_proxy), there is no change to the user interface, and I think it would be easier to work with.
Let me know what you think.
edit: change would look like this (moved PR to aleatory)
Hi,
I'm working on a neural proxy function for HMCECS and have a Taylor expansion proxy with an approximate Hessian. However, the file is becoming somewhat unruly as the proxies are currently in
hmc_gibbs.py
. If I move (only) the proxy functions to a separate file undercontrib
and keep using the static method interface for HMCECS (i.e.,HMCECS.taylor_proxy
), there is no change to the user interface, and I think it would be easier to work with.Let me know what you think.
edit: change would look like this (moved PR to aleatory)