view article
Figure 4
Training stage workflow. The model receives a prompt and image, generates a mask, and iteratively updates parameters via SFT and RL to align the output with the ground truth.

IUCrJ
Volume 13| Part 6| November 2026| Pages 748-765
ISSN: 2052-2525
https://doi.org/10.1107/S2052252526009887