File size: 3,091 Bytes
765160f | 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 | """Command line interface for one-model Helixbench runs."""
from __future__ import annotations
import argparse
import json
from pathlib import Path
from . import __version__
from .inputs import COHORT_SHA256, frozen_rows, prepare, sha256, validate_prepared
from .pipeline import METRICS, run
from .report import aggregate, calculate
def parser() -> argparse.ArgumentParser:
command = argparse.ArgumentParser(prog="helixbench", description="Six fixed Helixbench metrics")
command.add_argument("--version", action="version", version=__version__)
sub = command.add_subparsers(dest="command", required=True)
p = sub.add_parser("prepare", help="Validate the 1015 generated/GT media and freeze manifests")
p.add_argument("--generated", type=Path, required=True)
p.add_argument("--gt", type=Path, help="Optional if using packaged GT features")
p.add_argument("--model-id", required=True)
p.add_argument("--run-dir", type=Path, required=True)
p = sub.add_parser("run", help="Run one or all six metric stages")
p.add_argument("--run-dir", type=Path, required=True)
p.add_argument("--metric", choices=("all", *METRICS), default="all")
p.add_argument("--deps-root", type=Path, required=True)
p.add_argument("--weights-root", type=Path, required=True)
p.add_argument("--torch-home", type=Path, required=True)
p.add_argument("--frame-workers", type=int, default=8)
p = sub.add_parser("aggregate", help="Check coverage and emit exactly six scores")
p.add_argument("--run-dir", type=Path, required=True)
p = sub.add_parser("verify", help="Recalculate an existing six-score report")
p.add_argument("--run-dir", type=Path, required=True)
sub.add_parser("verify-package", help="Verify the frozen cohort and packaged engine hashes")
return command
def main() -> None:
args = parser().parse_args()
if args.command == "prepare":
print(json.dumps(prepare(args.generated, args.gt, args.run_dir, args.model_id), ensure_ascii=False))
elif args.command == "run":
run(args.run_dir, args.metric, args.deps_root, args.weights_root, args.torch_home, args.frame_workers)
elif args.command == "aggregate":
report = aggregate(args.run_dir)
print(json.dumps({key: value["value"] for key, value in report["metrics"].items()}, ensure_ascii=False))
elif args.command == "verify":
expected = json.loads((args.run_dir / "REPORT.json").read_text())
actual = calculate(args.run_dir)
if expected != actual:
raise ValueError("Stored report differs from recalculated six-metric result")
complete = json.loads((args.run_dir / "COMPLETE.json").read_text())
if complete["report_sha256"] != sha256(args.run_dir / "REPORT.json"):
raise ValueError("REPORT.json SHA256 differs from COMPLETE.json")
print("Helixbench report verified")
elif args.command == "verify-package":
from .provenance import verify_package
print(json.dumps(verify_package(), ensure_ascii=False))
if __name__ == "__main__":
main()
|