lutzroeder/netron · error · Error

Undefined node type.

Error message

Undefined node type.

What it means

coreml.Node's constructor requires obj.type to be truthy; a neural network layer dictionary without a 'type' key is considered invalid. Every CoreML layer must declare its layer type (e.g. 'convolution', 'pooling').

Source

Thrown at source/coreml.js:297

coreml.Value = class {

    constructor(name, type, description = null, initializer = null) {
        if (typeof name !== 'string') {
            throw new coreml.Error(`Invalid value identifier '${JSON.stringify(name)}'.`);
        }
        this.name = name;
        this.type = !type && initializer ? initializer.type : type;
        this.description = description;
        this.initializer = initializer;
        this.quantization = initializer ? initializer.quantization : null;
    }
};

coreml.Node = class {

    constructor(context, obj) {
        if (!obj.type) {
            throw new Error('Undefined node type.');
        }
        if (obj.group) {
            this.group = obj.group || null;
        }
        const type = context.metadata.type(obj.type);
        this.type = type ? { ...type } : { name: obj.type };
        this.type.name = obj.type.split(':').pop();
        this.name = obj.name || '';
        this.description = obj.description || '';
        this.inputs = (obj.inputs || []).map((argument) => {
            const values = argument.value.map((value) => value.value);
            return new coreml.Argument(argument.name, values, null, argument.visible);
        });
        this.outputs = (obj.outputs || []).map((argument) => {
            const values = argument.value.map((value) => value.value);
            return new coreml.Argument(argument.name, values, null, argument.visible);
        });
        this.attributes = Object.entries(obj.attributes || []).map(([name, value]) => {

View on GitHub (pinned to d8a543f5f8)

Solutions

1. Update netron to a version supporting your CoreML spec version
2. Inspect the model with coremltools to confirm layers are valid
3. Re-save/convert the model with current coremltools
4. If the layer is genuinely type-less, report the model as a netron issue

const layers = spec?.neuralNetwork?.layers ?? [];
if (layers.some((l) => !l.type)) throw new Error('model has untyped layers');

const hasType = (l) => Boolean(l && typeof l.type === 'string' && l.type.length);

try { model = new coreml.Model(...); } catch (e) { if (/Undefined node type/.test(e.message)) { /* skip or report the bad layer */ } }

When it happens

Trigger: Decoding a CoreML .mlmodel whose NeuralNetwork.layers array contains an entry with no/empty 'type' key; often when the FeatureType/protobuf mapping produced an unexpected object shape.

Common situations: Newer CoreML layer kinds not mapped by the installed netron version; corrupted or hand-constructed mlmodel; wrong protobuf branch decoded as layers.

Related errors

• Undefined node type.

AI-assisted analysis of lutzroeder/netron@d8a543f5f8 (2026-08-27). Data as JSON: /api/errors/bc09f3228132b66d. Report an issue: GitHub.