ruvnet/RuView · error · SystemExit

weights file not found: {weights_path}

Error message

weights file not found: {weights_path}

What it means

scripts/export-onnx.py converts the PoseNet safetensors checkpoint to ONNX. It takes the weights path from sys.argv[1], defaulting to ./pose_v1.safetensors relative to the current working directory, and raises SystemExit with this message when the path does not exist. In this repository the trained checkpoint actually lives at v2/crates/cog-pose-estimation/cog/artifacts/pose_v1.safetensors, so running the script from the repo root without arguments fails.

Source

Thrown at scripts/export-onnx.py:72

            if name == "__metadata__":
                continue
            start, end = meta["data_offsets"]
            shape = meta["shape"]
            dtype = meta["dtype"]
            assert dtype == "F32", f"unsupported dtype {dtype} for {name}"
            f.seek(8 + header_len + start)
            buf = f.read(end - start)
            arr = np.frombuffer(buf, dtype=np.float32).copy().reshape(shape)
            out[name] = torch.from_numpy(arr)
    return out


def main() -> None:
    weights_path = Path(sys.argv[1]) if len(sys.argv) > 1 else Path("pose_v1.safetensors")
    out_path = Path(sys.argv[2]) if len(sys.argv) > 2 else Path("pose_v1.onnx")

    if not weights_path.exists():
        raise SystemExit(f"weights file not found: {weights_path}")

    print(f"reading {weights_path}")
    tensors = load_safetensors(weights_path)
    print(f"  found {len(tensors)} tensors: {sorted(tensors.keys())}")

    model = PoseNet()
    # Map safetensors names (enc.c1.weight, head.fc1.weight, ...) to module params
    mapping = {
        "enc.c1.weight": "c1.weight",
        "enc.c1.bias": "c1.bias",
        "enc.c2.weight": "c2.weight",
        "enc.c2.bias": "c2.bias",
        "enc.c3.weight": "c3.weight",
        "enc.c3.bias": "c3.bias",
        "head.fc1.weight": "fc1.weight",
        "head.fc1.bias": "fc1.bias",
        "head.fc2.weight": "fc2.weight",
        "head.fc2.bias": "fc2.bias",

View on GitHub (pinned to 4685618388)

Solutions

1. Pass the checkpoint explicitly: `python scripts/export-onnx.py v2/crates/cog-pose-estimation/cog/artifacts/pose_v1.safetensors pose_v1.onnx`
2. Or run the script from the directory containing the weights so the default resolves
3. Locate the file first: `find . -name 'pose_v1.safetensors'` if you are unsure where the artifact is
4. If no checkpoint exists, produce or fetch the training artifact before exporting (the script cannot start from random init)

Example fix

# before
python scripts/export-onnx.py  # SystemExit: weights file not found: pose_v1.safetensors

# after
python scripts/export-onnx.py \
  v2/crates/cog-pose-estimation/cog/artifacts/pose_v1.safetensors \
  v2/crates/cog-pose-estimation/cog/artifacts/pose_v1.onnx

from pathlib import Path
weights = Path("v2/crates/cog-pose-estimation/cog/artifacts/pose_v1.safetensors")
assert weights.exists(), f"checkpoint missing: {weights}"

When it happens

Trigger: Running `python scripts/export-onnx.py` from a directory that does not contain pose_v1.safetensors; passing a relative weights path from a different cwd; typo or stale path after the artifacts moved.

Common situations: Executing the script from the repo root instead of the artifacts directory; a fresh clone where the training artifact has not been produced/downloaded yet; CI job that assumes cwd-relative defaults.

Related errors

• manifest not found: {MANIFEST_PATH}
• Calibration bundle {path} missing key {key!r}
• manifest is not valid JSON: {e}
• manifest must be an object with a 'markers' array
• duplicate marker ids: {sorted(dupes)}

AI-assisted analysis of ruvnet/RuView@4685618388 (2026-08-16). Data as JSON: /api/errors/61d08c6620fdce7a. Report an issue: GitHub.