keras-team/keras · error · ValueError

Layers added to a Sequential model can only have a single re

Error message

Layers added to a Sequential model can only have a single required positional argument, the input tensor. Layer {layer.__class__.__name__} has multiple required positional arguments: {required_positional_args}

What it means

Keras 3 Sequential inspects call() and requires exactly one required positional argument (the input tensor). A layer with multiple required positional arguments cannot be chained automatically, so build fails.

Source

Thrown at keras/src/models/sequential.py:225

                        inspect.Parameter.POSITIONAL_ONLY,
                        inspect.Parameter.POSITIONAL_OR_KEYWORD,
                    )
                ]
                required_positional_args = [
                    param
                    for param in positional_args
                    if param.default == inspect.Parameter.empty
                ]
                if not positional_args:
                    raise ValueError(
                        "Layers added to a Sequential model should "
                        "have a single positional argument, the "
                        "input tensor. Layer "
                        f"{layer.__class__.__name__} has no "
                        "positional arguments."
                    )
                if len(required_positional_args) > 1:
                    raise ValueError(
                        "Layers added to a Sequential model can "
                        "only have a single required positional "
                        "argument, the input tensor. Layer "
                        f"{layer.__class__.__name__} has multiple "
                        "required positional arguments: "
                        f"{required_positional_args}"
                    )
                raise e
        outputs = x
        self._functional = Functional(inputs=inputs, outputs=outputs)

    def call(self, inputs, training=None, mask=None):
        if self._functional:
            return self._functional.call(inputs, training=training, mask=mask)

        # Fallback: Just apply the layer sequence.
        # This typically happens if `inputs` is a nested struct.
        for layer in self.layers:

View on GitHub (pinned to 7a34a03db6)

Solutions

1. Move extra required arguments to __init__ or give them defaults in call()
2. Wrap the layer in a Lambda that closes over the extra values
3. Subclass Layer with call(self, inputs) only

Example fix

# before
class MyLayer(keras.layers.Layer):
    def call(self, inputs, scale): ...

# after
class MyLayer(keras.layers.Layer):
    def __init__(self, scale):
        super().__init__()
        self.scale = scale
    def call(self, inputs): ...

import inspect
sig = inspect.signature(layer.call)
req = [n for n, p in sig.parameters.items()
       if p.default is inspect.Parameter.empty
       and p.kind in (p.POSITIONAL_ONLY, p.POSITIONAL_OR_KEYWORD)]
assert len(req) <= 1, f'{layer.name} requires {req}'

When it happens

Trigger: class L(keras.layers.Layer): def call(self, inputs, bias): ... — then model.add(L()) and build

Common situations: Custom layers written with extra required hyperparameters in call(), or partial application of JAX/Flax style modules

Related errors

• Layers added to a Sequential model should have a single posi
• Cannot add call-context args after the layer has been called
• `call_function` argument is not supported with Sequential mo
• Expected `model` argument to be a `Sequential` model instanc
• Argument `input_tensors` must contain a single tensor.

AI-assisted analysis of keras-team/keras@7a34a03db6 (2026-08-25). Data as JSON: /api/errors/16a65e13df6b9a21. Report an issue: GitHub.