I am currently updating pymc3 to pymc, when using pm.sample_smc(XXX, nsteps = 50), it turns out that TypeError: SMC_KERNEL.init() got an unexpected keyword argument ‘nsteps’.
May I know how to get this solved? Thanks!
From the docs, it seems like it should be n_steps
pymc.smc.sample_smc — PyMC 4.0.0 documentation
It seems like the parameter was removed here in favor of correlation_threshold
. We have to update the docs! CC @aloctavodia
Sorry I do not quite get it, may I know how to fix it in the current code?
You should use correlation_threshold
instead.
I will update the docs.
Just to provide a short summary of the change. The number of steps is now computed automatically and dynamically at each stage. In most cases there is no need to change the value of the correlation_threshold argument.
In that case, may I know if there is still way to assess the n_steps at each stage? Thanks !
I think they are recorded in the sampling stats when the final results are returned.