Extracting warmup samples in turing.jl

Hello,

Im running inference for my model in turing.jl. Im using NUTS() sampler where I give some warmup. I was wondering that if I can extract these warmup samples ? rather than getting chain plot with just samples after warmup.

Thank you

Hey! Sorry I don’t have a direct answer but I noticed that this tutorial (Bayesian Poisson Regression – Turing.jl) goes into the topic. I think using discard_adapt = false within the sample method and then chains[1:N, :, :] to get the N adaptation iterations might get you there.

1 Like