model card: shorter, author name

model card: simplify, single author

model card: describe v4 only, drop the v1/v2 comparison

v4 replaces v2: root weights removed (v2 kept under v2/)

v2: move the v2 weights into v2/ (root copy kept until the Space is on v4)

v4: add README.md

v4: add v4/lgbm_s7.joblib

v4: add v4/lgbm_s1.joblib

v4: add v4/lgbm_s42.joblib

v4: add v4/meta.json

v4: add v4/backbone.pt

v2 at root, v1 in v1/; simplify the model and performance sections

Remove the now-duplicated v2/ folder

Move v2 to the repo root and v1 into v1/

Drop hand-crafted from the feature description

Document v1 and v2; replace the pre-leak-fix accuracy table with verified v2 numbers

Add v2 weights: 96 morphometric + 768 fine-tuned DINOv2 ViT-B/14

Upload README.md with huggingface_hub

Upload feature_scaler_RandomForest.joblib with huggingface_hub

Upload model_RandomForest.joblib with huggingface_hub

Upload backbone.pt with huggingface_hub

Upload feature_scaler_LightGBM.joblib with huggingface_hub

Upload model_LightGBM.joblib with huggingface_hub

Upload feature_scaler_XGBoost.joblib with huggingface_hub

Upload model_XGBoost.joblib with huggingface_hub

Upload label_encoder.joblib with huggingface_hub

Upload feature_columns.joblib with huggingface_hub

initial commit