docs: restyle model card to match code-release README

docs: RAE encoder/decoder provenance

docs: citation + acknowledgments

docs: drop discriminators row

remove discriminators/ (stage-1 retraining not a supported use case; pretrained decoders are provided)

docs: clarify discriminators/ is stage-1-retraining only

docs: clearer folder names, code-repo URL, fid_reference note

add fid_reference/VIRTUAL_imagenet256_labeled.npz (from openai/guided-diffusion)

rename: stats->normalization_stats, discs->discriminators, guides->autoguidance

add README.md

add stats/mae-b/stat.pt

add stats/siglip2-b/stat.pt

add stats/dinov2-b/stat.pt

add decoders/mae-b/ViTXL_n08/model.pt

add decoders/siglip2-b/ViTXL_n08/model.pt

add decoders/dinov2-b/ViTXL_n08/model.pt

add gmm/mae-b/spatial_gmm_from_cls_n8192_diag_k-means++_flip.pkl

add gmm/mae-b/gmm_n8192_diag_k-means++.pkl

add gmm/siglip2-b/spatial_gmm_from_cls_n8192_diag_k-means++_flip.pkl

add gmm/siglip2-b/gmm_n8192_diag_k-means++.pkl

add gmm/dinov2-b/spatial_gmm_from_cls_n8192_diag_k-means++_flip.pkl

add gmm/dinov2-b/gmm_n8192_diag_k-means++.pkl

add checkpoints/guides/dit-s-mode-gmm-25k.pt

add checkpoints/guides/dit-s-uncond-gmm-25k.pt

add checkpoints/mae-b/mode-gmm-25k.pt

add checkpoints/mae-b/uncond-gmm-25k.pt

add checkpoints/siglip2-b/mode-gmm-25k.pt

add checkpoints/siglip2-b/uncond-gmm-25k.pt

add checkpoints/dinov2-b/mode-gmm-100k.pt

add checkpoints/dinov2-b/mode-gmm-25k.pt

add checkpoints/dinov2-b/uncond-gmm-100k.pt

add discs/dino_vit_small_patch8_224.pth

add checkpoints/dinov2-b/uncond-gmm-25k.pt

initial commit