huggingface/pytorch-image-models · error · ValueError

Preset '{value}' is empty or invalid

Error message

Preset '{value}' is empty or invalid

What it means

Thrown by resolve_ns_coefficients in timm's Muon optimizer when a named Nesterov-style coefficients preset exists in the presets dict but maps to an empty or non-sequence value. It is a data-integrity check on the internal presets table, guarding before attempting to iterate the preset's items.

Source

Thrown at timm/optim/muon.py:1035

        presets: Mapping[str, Sequence[Sequence[float]]]
) -> List[Tuple[float, float, float]]:
    # tiny helpers (kept inline for succinctness)
    is_seq = lambda x: isinstance(x, Sequence) and not isinstance(x, (str, bytes))
    is_real = lambda x: isinstance(x, numbers.Real) and not isinstance(x, bool)

    def as_coeff(x: Sequence[float]) -> Tuple[float, float, float]:
        if not is_seq(x) or len(x) != 3 or not all(is_real(v) for v in x):
            raise ValueError(f"Coefficient must be length-3 of real numbers, got: {x!r}")
        a, b, c = x  # type: ignore[misc]
        return float(a), float(b), float(c)

    if isinstance(value, str):
        if value not in presets:
            valid = ", ".join(sorted(presets.keys()))
            raise ValueError(f"Unknown coefficients preset '{value}'. Valid options: {valid}")
        seq = presets[value]
        if not is_seq(seq) or len(seq) == 0:
            raise ValueError(f"Preset '{value}' is empty or invalid")
        return [as_coeff(item) for item in seq]  # validate & cast

    if not is_seq(value):
        raise TypeError(
            "Coefficients must be a preset name (str), a 3-sequence (a,b,c), "
            "or a sequence of 3-sequences."
        )

    # Decide single triple vs list-of-triples by structure
    if len(value) == 3 and all(is_real(v) for v in value):  # type: ignore[index]
        return [as_coeff(value)]  # single triple -> wrap

    # Otherwise treat as list/tuple of triples
    out = []
    for i, item in enumerate(value):  # type: ignore[assignment]
        if not is_seq(item):
            raise TypeError(f"Item {i} is not a sequence: {item!r}")
        out.append(as_coeff(item))

View on GitHub (pinned to 9a5261e31b)

Solutions

1. Use a well-known built-in preset name (print/inspect the presets dict to see valid entries)
2. If defining custom presets, ensure each value is a non-empty sequence of 3-sequences
3. Pass explicit coefficients as a (a,b,c) tuple or list of triples instead of a preset name

Example fix

# before
opt = Muon(params, ns_coefficients='my_preset')  # preset entry is []

# after
opt = Muon(params, ns_coefficients=[(0.1, 0.5, 0.9)])

from timm.optim.muon import NS_COEFFICIENT_PRESETS  # inspect presets dict
name = 'my_preset'
assert name in NS_COEFFICIENT_PRESETS and len(NS_COEFFICIENT_PRESETS[name]) > 0

def valid_preset(name: str, presets: dict) -> bool:
    seq = presets.get(name)
    return isinstance(seq, (list, tuple)) and len(seq) > 0

When it happens

Trigger: Passing ns_coefficients='preset_name' to timm.optim.Muon/AdamMuon where the presets dict entry for that name is empty ([]) or not a sequence (e.g. None or a scalar).

Common situations: Custom/monkeypatched presets tables, stale forks where a preset was emptied, or programmatic preset generation that produced an empty list.

Related errors

• Coefficient list cannot be empty
• Coefficients must be a preset name (str), a 3-sequence (a,b,
• Item {i} is not a sequence: {item!r}
• Invalid learning rate: {}
• Invalid learning rate: {lr}

AI-assisted analysis of huggingface/pytorch-image-models@9a5261e31b (2026-08-27). Data as JSON: /api/errors/5374db419e87d6e7. Report an issue: GitHub.