view article

Figure 6
Negative cross-validation log likelihood reduction during hyperparameter optimization for the simulated grain presented in Figs. 4[link] and 5[link]. Optimization was conducted using the L-BFGS-B algorithm as implemented in SciPy with a maximum of ten line-search steps per iteration. Gradients were computed using automatic differentiation as implemented in PyTorch.

Journal logoJOURNAL OF
ISSN: 1600-5767
Follow J. Appl. Cryst.
Sign up for e-alerts
Follow J. Appl. Cryst. on Twitter
Follow us on facebook
Sign up for RSS feeds