-
Notifications
You must be signed in to change notification settings - Fork 18
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
sample
which takes in an initial state
#109
Comments
I think it is a useful feature for all samplers, which would fit nicely here. |
To slightly widen the scope of this issue, we can also consider saving/restoring sampler states (e.g. step-size and preconditioning matrix of HMC). |
Saving in what sense? This would all be included in the |
Yes, I think that's one of the things that could/should be moved from DynamicPPL/Turing to AbstractMCMC (maybe I misremember but it feels like there was also some discussion in some issue or PR a while ago?). |
It would be nice to move have something similar to
resume_from
that we have in DPPL (https://github.com/TuringLang/DynamicPPL.jl/blob/217b8cc3a0fbee96049b90a53f5db9f2a586ba96/src/sampler.jl#L84-L95) in AbstractMCMC.In particular, by default, it shouldn't be an issue to just make
assuming that the
resume_from
corresponds to the samplerstate
.Thoughts @devmotion @cpfiffer @yebai ?
CC: @HarrisonWilde
The text was updated successfully, but these errors were encountered: