I am currently trying to use a PyMC3 model in an online learning setting. The data arrives in single instances and I need to update the model on the complete data in each iteration. Since the posterior distribution only changes slightly and I want to avoid having to wait for the tuning steps to com…

Reuse tuning for next sampling call

twiecki February 4, 2019, 3:21pm 3

Couldn’t you also just recycle the potential from a previous run NUTS object?

Topic		Replies	Views
Pause and resume training a model that can change during training Questions	2	1438	May 2, 2019
Restarting sampling from stored multitrace: mixed sampling and tuning Questions	2	586	October 5, 2020
Saving the state of the sampler in pymc3? Questions	11	1579	August 20, 2020
Reusing tuned NUTS steps to save time Questions	0	438	February 19, 2021
Just using NUTS Questions	7	1168	May 5, 2018