| |
| """Upload an already-complete local evaluation result to SwanLab.""" |
|
|
| from __future__ import annotations |
|
|
| import argparse |
| import json |
| from pathlib import Path |
|
|
| import swanlab |
|
|
|
|
| def main() -> None: |
| parser = argparse.ArgumentParser() |
| parser.add_argument("--result-dir", type=Path, required=True) |
| parser.add_argument("--run-name", required=True) |
| parser.add_argument("--project", default="unified-model") |
| args = parser.parse_args() |
| result_dir = args.result_dir.resolve() |
| payload = json.loads((result_dir / "metrics.json").read_text(encoding="utf-8")) |
| metadata = payload["metadata"] |
| metrics = payload["metrics"] |
| benchmark = str(metadata["benchmark"]) |
| run = swanlab.init( |
| project=args.project, |
| name=args.run_name, |
| log_dir=str(result_dir / "swanlab_retry"), |
| config=metadata, |
| tags=[benchmark, "evaluation", "checkpoint-7000", "8GPU"], |
| description=f"Judge-free exact-match evaluation on {benchmark}; uploaded from validated local artifacts.", |
| ) |
| swanlab.log(metrics, step=int(metadata["checkpoint_step"])) |
| status = { |
| "enabled": True, |
| "retry_upload": True, |
| "run_id": getattr(run, "id", None), |
| "run_url": str(getattr(run, "url", "") or "") or None, |
| } |
| swanlab.finish() |
| (result_dir / "swanlab.json").write_text( |
| json.dumps(status, ensure_ascii=False, indent=2, sort_keys=True) + "\n", |
| encoding="utf-8", |
| ) |
| print(json.dumps(status, sort_keys=True)) |
|
|
|
|
| if __name__ == "__main__": |
| main() |
|
|
