Create scorer.py

scorer.py

@@ -0,0 +1,78 @@
+from dataclasses import dataclass
+from typing import Dict, Any, List
+import re
+
+REQ = [
+    "decoherence_onset_time",
+    "precursor_pattern_type",
+    "severity_gradient",
+    "failure_likelihood_index",
+    "estimated_time_to_critical_min",
+    "primary_decoupling_channels",
+]
+
+@dataclass
+class ScoreResult:
+    score: float
+    details: Dict[str, Any]
+
+def _time_min(p: str):
+    # accepts t=38m or 38m
+    m = re.search(r"decoherence_onset_time\s*[:=]\s*(t\s*=\s*)?([0-9]+(\.[0-9]+)?)\s*m", p)
+    if not m:
+        return None
+    return float(m.group(2))
+
+def _f(p: str, key: str):
+    m = re.search(rf"{key}\s*[:=]\s*(0\.\d+|1\.0)\b", p)
+    return float(m.group(1)) if m else None
+
+def _i(p: str, key: str):
+    m = re.search(rf"{key}\s*[:=]\s*(\d+)\b", p)
+    return int(m.group(1)) if m else None
+
+def _ttc(p: str):
+    m = re.search(r"estimated_time_to_critical_min\s*[:=]\s*(\d+)", p)
+    return int(m.group(1)) if m else None
+
+def score(sample: Dict[str, Any], prediction: str) -> ScoreResult:
+    p = (prediction or "").lower()
+    words_ok = len(p.split()) <= 1100
+
+    hits = sum(1 for k in REQ if k in p)
+
+    t = _time_min(p)
+    sev = _f(p, "severity_gradient")
+    lik = _f(p, "failure_likelihood_index")
+    ttc = _ttc(p)
+
+    numeric_ok = int(
+        t is not None and 0.0 <= t <= 10000.0 and
+        sev is not None and 0.0 <= sev <= 1.0 and
+        lik is not None and 0.0 <= lik <= 1.0 and
+        ttc is not None and 0 <= ttc <= 20000
+    )
+
+    type_ok = int("precursor_pattern_type" in p and len(p) > 70)
+    chan_ok = int("primary_decoupling_channels" in p and ("egt" in p or "n1" in p or "ff" in p))
+
+    # sanity: higher severity should not imply very low likelihood
+    sanity = 0
+    if sev is not None and lik is not None:
+        sanity = int(lik + 0.15 >= sev)
+
+    raw = (
+        0.15 * int(words_ok) +
+        0.40 * (hits / len(REQ)) +
+        0.20 * numeric_ok +
+        0.10 * type_ok +
+        0.10 * chan_ok +
+        0.05 * sanity
+    )
+
+    return ScoreResult(score=min(1.0, raw), details={"id": sample.get("id"), "hits": hits, "sanity": sanity})
+
+def aggregate(results: List[ScoreResult]) -> Dict[str, Any]:
+    if not results:
+        return {"mean": 0.0, "n": 0}
+    return {"mean": sum(r.score for r in results)/len(results), "n": len(results)}