beat
beat copied to clipboard
compilation time
Compilation time of the forward model seems to scale with the number of Markov Chains involved in the sampling. This shouldnt be like that!
might be related to theano amd linked library
resolved in BEAT v2, and migration to pytensor