huggingface/candle · error

missing input {}

Error message

missing input {}

What it means

Before running the graph, simple_eval_ iterates graph.input and looks up each declared input in the user-supplied values HashMap. If a name declared as a model input is absent from the inputs map, evaluation cannot proceed and this error is thrown. Input names in ONNX are exact-match strings.

Source

Thrown at candle-onnx/src/eval.rs:273

        let tensor = get_tensor(t, t.name.as_str())?;
        values.insert(t.name.to_string(), tensor);
    }
    for input in graph.input.iter() {
        let input_type = match &input.r#type {
            Some(input_type) => input_type,
            None => continue,
        };
        let input_type = match &input_type.value {
            Some(input_type) => input_type,
            None => continue,
        };
        let tensor_type = match input_type {
            onnx::type_proto::Value::TensorType(tt) => tt,
            _ => continue,
        };

        let tensor = match values.get(&input.name) {
            None => bail!("missing input {}", input.name),
            Some(tensor) => tensor,
        };
        let dt = match DataType::try_from(tensor_type.elem_type) {
            Ok(dt) => match dtype(dt) {
                Some(dt) => dt,
                None => {
                    bail!("unsupported 'value' data-type {dt:?} for {}", input.name)
                }
            },
            type_ => bail!("unsupported input type {type_:?}"),
        };
        match &tensor_type.shape {
            None => continue,
            Some(shape) => {
                if shape.dim.len() != tensor.rank() {
                    bail!(
                        "unexpected rank for {}, got {:?}, expected {:?}",
                        input.name,

View on GitHub (pinned to d5fee525bf)

Solutions

1. Print model.graph.input names and ensure every one has a matching key in the inputs HashMap (exact string match, case-sensitive).
2. Check whether the missing name is actually an initializer/parameter rather than a runtime input; those must not be supplied.
3. Use the model's exported input signature from the exporter (e.g. torch.onnx.export input_names) to build the map.
4. Handle the error and report which input name is required in your app's validation layer.

Example fix

// before
let mut inputs = HashMap::new();
inputs.insert("x".to_string(), xs);
// after
let mut inputs = HashMap::new();
inputs.insert("input".to_string(), xs); // name from model.graph.input
inputs.insert("attention_mask".to_string(), mask);

fn missing_inputs(model: &onnx::ModelProto, inputs: &HashMap<String, Value>) -> Vec<String> {
    model.graph.as_ref().map_or(vec![], |g| g.input.iter()
        .filter(|i| !inputs.contains_key(&i.name))
        .map(|i| i.name.clone()).collect())
}

if let Err(e) = simple_eval(&model, inputs) {
    if let Some(name) = e.to_string().strip_prefix("missing input ") {
        anyhow::bail!("model requires input '{name}' — check spelling/casing");
    }
    return Err(e.into());
}

When it happens

Trigger: simple_eval called with an inputs HashMap missing one or more names declared in model.graph.input — typically a typo'd key, a case mismatch, or forgetting a required input (ONNX models have multiple inputs).

Common situations: Multi-input models (e.g. tokens + attention_mask); inputs accidentally fed only as initializers; renaming of inputs between exporter versions; passing input named 'x' when the model declares 'input' or 'input_ids'.

Related errors

• unexpected rank for {}, got {:?}, expected {:?}
• unexpected dtype for {}, got {:?}, expected {dt:?}
• cannot find {input_name} for op '{}'
• unsupported 'value' data-type {dt:?} for {}
• unsupported input type {type_:?}

AI-assisted analysis of huggingface/candle@d5fee525bf (2026-09-02). Data as JSON: /api/errors/cc530cb924ebd233. Report an issue: GitHub.