Closed dp-rice closed 1 year ago
For some reason, I'm not seeing this anymore. At least fits.tsv
and fits_summary.tsv
seem to be deterministic. It could be that it was always deterministic and the input was changing without me realizing it. It could be that we're not saving the non-deterministic stuff in the repo anymore.
We're providing a random seed, but the Stan output is not deterministic. Maybe the seed is only used for the HMC sampler and we need to pass a seed explicitly to the RNG part of the code.