Datasets:
File size: 4,117 Bytes
152a1c3 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 | #!/usr/bin/env python3
"""Score a prediction file against the labels for one property and condition."""
import argparse
import json
import sys
from pathlib import Path
RELEASE = Path(__file__).resolve().parents[1]
DATA = RELEASE / "data"
PROPERTIES = (
"dipole",
"gap",
"infrared",
"raman",
"ultraviolet_visible",
"proton_nmr_peaks",
"proton_nmr_integration",
"proton_nmr_atom",
"carbon_nmr_peaks",
"carbon_nmr_degeneracy",
"carbon_nmr_atom",
"electronic",
"magnetic_resonance",
"vibrational",
"all_properties",
)
CONDITIONS = ("correct", "shuffled", "counterfactual")
def load_jsonl(path):
rows = []
for line_number, line in enumerate(path.read_text().splitlines(), start=1):
if not line.strip():
continue
try:
row = json.loads(line)
except json.JSONDecodeError as error:
raise ValueError(f"{path}:{line_number}: {error}") from error
if not isinstance(row, dict) or "record_id" not in row:
raise ValueError(f"{path}:{line_number}: missing record_id")
rows.append(row)
return rows
def load_task(property_name, condition):
if property_name not in PROPERTIES or condition not in CONDITIONS:
raise ValueError(f"unknown property or condition: {property_name} {condition}")
tasks = load_jsonl(DATA / property_name / f"{condition}.jsonl")
labels = load_jsonl(DATA / property_name / f"{condition}.labels.jsonl")
if [row["record_id"] for row in tasks] != [row["record_id"] for row in labels]:
raise ValueError(f"task and label record_id lists differ: {property_name} {condition}")
scored = []
for task, label in zip(tasks, labels):
if not isinstance(label.get("label"), int):
raise ValueError(f"label is not an integer: {label['record_id']}")
n_candidates = len(task["candidate_completed_smiles"])
if not 0 <= label["label"] < n_candidates:
raise ValueError(f"label out of range: {label['record_id']}")
scored.append((task["record_id"], label["label"], n_candidates))
return scored
def load_predictions(path):
chosen = {}
for row in load_jsonl(path):
record_id = row["record_id"]
if record_id in chosen:
raise ValueError(f"duplicate prediction: {record_id}")
if row.get("invalid") is True or "selected_candidate_index" not in row:
chosen[record_id] = None
continue
index = row["selected_candidate_index"]
if not isinstance(index, int) or isinstance(index, bool):
raise ValueError(f"selected_candidate_index is not an integer: {record_id}")
chosen[record_id] = index
return chosen
def score(property_name, condition, predictions):
hits = 0
invalid = 0
missing = 0
out_of_range = 0
tasks = load_task(property_name, condition)
for record_id, label, n_candidates in tasks:
if record_id not in predictions:
missing += 1
continue
choice = predictions[record_id]
if choice is None:
invalid += 1
continue
if not 0 <= choice < n_candidates:
out_of_range += 1
continue
hits += choice == label
n = len(tasks)
return {
"property": property_name,
"condition": condition,
"n": n,
"hits": hits,
"accuracy": hits / n if n else None,
"missing": missing,
"invalid": invalid,
"out_of_range": out_of_range,
}
def main():
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument("--property", required=True, choices=PROPERTIES)
parser.add_argument("--condition", required=True, choices=CONDITIONS)
parser.add_argument("--predictions", required=True, type=Path)
args = parser.parse_args()
result = score(args.property, args.condition, load_predictions(args.predictions))
json.dump(result, sys.stdout, indent=2)
sys.stdout.write("\n")
return 0
if __name__ == "__main__":
raise SystemExit(main())
|