The released weights contain one checkpoint, whose predictor is (1, 2048, 20484) — the cortical model. Section 5.6 describes a separately trained subcortical model over 8,802 Harvard-Oxford voxels in 8 regions, and tribev2/grids/run_subcortical.py publishes the grid for it (MaskProjector(mask="subcortical"), fwhm 6.0). Only the weights appear to be missing.
Would you consider releasing the subcortical checkpoint as well?
The released weights contain one checkpoint, whose predictor is
(1, 2048, 20484)— the cortical model. Section 5.6 describes a separately trained subcortical model over 8,802 Harvard-Oxford voxels in 8 regions, andtribev2/grids/run_subcortical.pypublishes the grid for it (MaskProjector(mask="subcortical"),fwhm 6.0). Only the weights appear to be missing.Would you consider releasing the subcortical checkpoint as well?