NUTS speed is very slow for high dimension parameter inference in Turing.jl

[danielw2904]

What about Turing.setrdcache(true) in this case?

This worked well for me in a different situation:

Turing.jl for Causal Inference model

Here is my benchmark for two runs of each: Turing.setadbackend(:forwarddiff) Iterations = 1001:1:2000 Number of chains = 4 Samples per chain = 1000 Wall duration = 102.14 seconds Compute duration = 361.99 seconds Iterations = 1001:1:2000 Number of chains = 4 Samples per chain = 1000 Wall duration = 95.9 seconds Compute duration = 333.32 seconds Turing.setadbackend(:tracker) Iterations = 1001:1:2000 Number of chains = 4 Samples per chain = 1000 Wall duration…