Name every model's source and terms, and carry those licenses in full

Script the DINOv2 wrapper so it keeps its defaults and reads the statistics it is passed

Add the anatomix ViT encoder, trace both anatomix networks in evaluation mode

Floor the MIND variance so a constant batch yields a flat descriptor

test one file

TS models: rebuild with fixed Canonical (device-safe tensor, drop aten::dim); weights bit-identical

Expand models.json to all 30 models (per-directory FOV/dim/channels)

Add models.json registry: per-model intrinsics (dimension, channels, field-of-view) for IMPACT registration presets

Fix MIND export: conv padding=0 (match MINDSSC) + fuser-safe Normalize (stats.numel)

Remove clipping in CT model for IMPACT-synth

Add VGG16 feature extractor for ImpactSynth baseline

Move input normalization to models and make layer selection part of forward()

Add M252 and M253

Add MRSegmentator weights, TS/M850 and SAM2.1 Small

Update README.md

Update README.md

Update README

Add github link

Initial upload of TorchScript models for IMPACT

Initial upload of TorchScript models for IMPACT

initial commit