| """Reference model for X-ROBUST-009: TF-IDF + logistic regression with temperature scaling; the confidence is the | |
| calibrated maximum class probability and the model abstains below a threshold fitted on validation data.""" | |
| import json | |
| import sys | |
| from pathlib import Path | |
| import numpy as np | |
| sys.path.insert(0, str(Path(__file__).resolve().parent)) | |
| from src import features as F # noqa: E402 | |
| def load(checkpoint_dir, models_dir, device): | |
| ck = Path(checkpoint_dir) | |
| cal = json.loads((ck / "calibration.json").read_text(encoding="utf-8")) | |
| return {"vecs": F.load(ck), "W": np.load(ck / "coef.npy").astype(np.float64), | |
| "b": np.load(ck / "intercept.npy").astype(np.float64), **cal} | |
| def predict(state, texts): | |
| z = (F.transform(state["vecs"], texts) @ state["W"].T + state["b"]) / state["temperature"] | |
| z = z - z.max(1, keepdims=True) | |
| p = np.exp(z) | |
| p /= p.sum(1, keepdims=True) | |
| conf = p.max(1) | |
| return p.argmax(1), conf, conf < state["threshold"] | |