| """Load the compact MetNet-3 checkpoint and run fake-data inference.""" |
|
|
| import json |
| import sys |
| from pathlib import Path |
|
|
| import torch |
|
|
|
|
| ROOT = Path(__file__).resolve().parents[1] |
| sys.path.insert(0, str(ROOT)) |
| sys.path.insert(0, str(Path(__file__).resolve().parent)) |
|
|
| from model import MetNet3, MetNet3Config |
| from model.fake_data import make_fake |
|
|
|
|
| def main() -> None: |
| result_dir = ROOT / "result" |
| result_dir.mkdir(parents=True, exist_ok=True) |
| checkpoint = torch.load(ROOT / "weight" / "model.pth", map_location="cpu", weights_only=True) |
| config = MetNet3Config(**checkpoint["config"]) |
| model = MetNet3(config) |
| model.load_state_dict(checkpoint["model"]) |
| model.eval() |
| batch, targets = make_fake(config) |
| with torch.inference_mode(): |
| outputs = model(batch) |
| torch.save(outputs, result_dir / "prediction.pt") |
| torch.save(targets, result_dir / "target.pt") |
| summary = {name: list(value.shape) for name, value in outputs.items()} |
| summary["finite"] = all(bool(torch.isfinite(value).all()) for value in outputs.values()) |
| (result_dir / "inference.json").write_text(json.dumps(summary, indent=2) + "\n") |
| print(json.dumps(summary, indent=2)) |
|
|
|
|
| if __name__ == "__main__": |
| main() |
|
|