Vision Rules demo assets — built by scripts/vision_pack_assets.py
from get_concept_into_rules/docker_demo/app/runs/celeba/efficientnet_b0_sequential/exp_4
(EfficientNet-B0 concept bottleneck over the 38 CelebA attributes,
female/male head; TT-Sparse rulebooks cut at the training median).
Rulebooks: compact = ttsparse_30n5b_median_drop0.5.json (6 rules), accurate = ttsparse_30n5b_median.json (26 rules).
The checkpoint keeps only backbone.* and concept_head.* (the label head
is unused: the verdict is the rulebook's). Gallery photos are the demo's
test faces, EXIF-transposed and downscaled to 640 px.
Gallery ids: carla, famous, haaland, harry, mbappe, pam, strangerthings
Regenerate: python scripts/vision_pack_assets.py <docker_demo/app> src/implicant/vision/assets
