keras-team/keras · error · ValueError

Theta of a Thresholded ReLU layer cannot be None, expecting

Error message

Theta of a Thresholded ReLU layer cannot be None, expecting a float. Received: {theta}

What it means

The deprecated ThresholdedReLU layer requires a numeric `theta` (the activation threshold); passing theta=None raises this ValueError because the constructor immediately converts theta to a tensor of the layer's compute dtype, which None cannot satisfy. The default is 1.0, so this fires only when None is passed explicitly — most often an unset config key forwarded verbatim.

Source

Thrown at keras/src/legacy/layers.py:222

    def get_config(self):
        config = {
            "factor": self.factor,
            "interpolation": self.interpolation,
            "seed": self.seed,
        }
        base_config = super().get_config()
        return {**base_config, **config}


@keras_export("keras._legacy.layers.ThresholdedReLU")
class ThresholdedReLU(Layer):
    """DEPRECATED."""

    def __init__(self, theta=1.0, **kwargs):
        super().__init__(**kwargs)
        if theta is None:
            raise ValueError(
                "Theta of a Thresholded ReLU layer cannot be None, expecting a "
                f"float. Received: {theta}"
            )
        if theta < 0:
            raise ValueError(
                "The theta value of a Thresholded ReLU layer "
                f"should be >=0. Received: {theta}"
            )
        self.supports_masking = True
        self.theta = tf.convert_to_tensor(theta, dtype=self.compute_dtype)

    def call(self, inputs):
        dtype = self.compute_dtype
        return inputs * tf.cast(tf.greater(inputs, self.theta), dtype)

    def get_config(self):
        config = {"theta": float(self.theta)}
        base_config = super().get_config()

View on GitHub (pinned to 7a34a03db6)

Solutions

1. Omit theta entirely to use the default 1.0, or pass an explicit float
2. Replace config.get('theta') with config.get('theta', 1.0)
3. If theta is computed dynamically, assert it is numeric before constructing the layer

Example fix

# before
theta = cfg.get('theta')  # None when missing
layer = ThresholdedReLU(theta=theta)
# after
theta = cfg.get('theta', 1.0)
layer = ThresholdedReLU(theta=theta)

theta = cfg.get('theta', 1.0)
assert isinstance(theta, (int, float)), f'theta must be a float, got {theta!r}'

def is_valid_theta(t) -> bool:
    return isinstance(t, (int, float)) and not isinstance(t, bool)

When it happens

Trigger: Constructing keras._legacy.layers.ThresholdedReLU(theta=None) — typically because a config dict/YAML omitted or nullified theta and the value was forwarded via .get('theta').

Common situations: Building layers from hyperparameter dicts where dict.get('theta') yields None; deserializing configs written for a different layer where theta is absent; config merges where None overrides a default.

Understand the failure class

Background: "missing required argument" and "the following required arguments were not provided": what required-argument errors mean and how to fix them — this error's family across 20 libraries.

Related errors

• The theta value of a Thresholded ReLU layer should be >=0. R
• Expected `padding` to be a tuple of 3 tuples of 2 integers.
• Expected `padding` to be a tuple of 2 integers. Received: pa
• `factor` argument cannot have an upper bound lesser than the
• `factor` argument must have values larger than -1. Received:

AI-assisted analysis of keras-team/keras@7a34a03db6 (2026-08-25). Data as JSON: /api/errors/8297940446801482. Report an issue: GitHub.