keras-team/keras · error · ValueError

axis size must be divisible by 2. Received: x.shape={x.shape

Error message

axis size must be divisible by 2. Received: x.shape={x.shape} with axis={axis}

What it means

Error "axis size must be divisible by 2. Received: x.shape={x.shape} with axis={axis}" thrown in keras-team/keras.

Source

Thrown at keras/src/backend/numpy/ops/nn.py:181

            * (scipy.special.erf(x / sqrt_2) + 1).astype(x.dtype)
            / np.array(2, x.dtype)
        )


def celu(x, alpha=1.0):
    x = convert_to_tensor(x)
    alpha = np.array(alpha, x.dtype)
    return np.maximum(x, np.array(0.0, dtype=x.dtype)) + alpha * np.expm1(
        np.minimum(x, np.array(0.0, dtype=x.dtype)) / alpha
    )


def glu(x, axis=-1):
    x = convert_to_tensor(x)
    dtype = x.dtype
    canonicalize_axis(axis, len(x.shape))
    if x.shape[axis] % 2 != 0:
        raise ValueError(
            "axis size must be divisible by 2. "
            f"Received: x.shape={x.shape} with axis={axis}"
        )
    x1, x2 = np.split(x, 2, axis)
    return (x1 * sigmoid(x2)).astype(dtype)


def hard_tanh(x):
    x = convert_to_tensor(x)
    min_val = np.asarray(-1.0, x.dtype)
    max_val = np.asarray(1.0, x.dtype)
    return np.array(np.clip(x, min_val, max_val), dtype=x.dtype)


def hard_shrink(x, threshold=0.5):
    x = convert_to_tensor(x)
    threshold = np.asarray(threshold, x.dtype)
    return np.array(

View on GitHub (pinned to 7a34a03db6)

When it happens

Trigger: Thrown at keras/src/backend/numpy/ops/nn.py:181 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of keras-team/keras@7a34a03db6 (2026-08-25). Data as JSON: /api/errors/1d6f9c7ef75a6339. Report an issue: GitHub.