Inferred model parameters versus input model parameters for the case in which spectra have between 10 000–100 000 counts, spread over 200 bins. A single round inference is performed on an initial sample of 10 000 simulated spectra. The median of the posterior distributions is computed from 20 000 samples and the error on the median is computed from the 68% quantile of the distribution. The linear regression coefficient is computed for each parameter over the 500 test samples.