Closed luiarthur closed 4 years ago
Looks like the computation of the effective sample size has been changed and doesn't work with the resulting chain. I can have a look. For now, you can run
chain = sample(model_fun, SMC(), iterations);
To ensure that the describe
function is not called. (We call it if Julia requests the show
method, i.e. the REPL shows the resulting chain.)
Thanks. So, I tried adding the semi-colon, and it ran because no output was printed, so no calls to describe
were made.
I'm able to call chain.value
, but chain[:mu]
, etc. throws essentially the same error as before.
I think I'm able to continue from here. Thanks!
Should I close this issue and create a separate issue on the Turing repo?
Yes, that is a good idea.
I am unsuccessful in replicating the infinite GMM example here: https://turing.ml/dev/tutorials/6-infinitemixturemodel/
I'm using the
Project.toml
andManifest.toml
files from: https://github.com/luiarthur/TuringBnpBenchmarks/This is what I'm trying to run. It's almost verbatim what's in the tutorial (just replaced some Unicode):
This is the error I'm getting: