Evaluation: replace threshold-based F1 tables with AUROC / 3-class F1 / binary F1 (uncertain→neg) on CT-RATE + SNUH, with 95% CIs; add ≥1-positive subset

Upload README.md with huggingface_hub

Upload modeling_chest2vec_labeler.py with huggingface_hub

Upload README.md with huggingface_hub

Upload modeling_chest2vec_labeler.py with huggingface_hub

Upload config.json with huggingface_hub

chest2vec CT report labeler (0.6B): self-contained AutoModel, 137-leaf ternary, CheXbert-style report F1

initial commit