keras-team/keras · error · NotImplementedError

Argument synchronized=True is not supported with NumPy.

Error message

Argument synchronized=True is not supported with NumPy.

What it means

Error "Argument synchronized=True is not supported with NumPy." thrown in keras-team/keras.

Source

Thrown at keras/src/backend/numpy/nn.py:936

    if target.shape != output.shape:
        raise ValueError(
            "Arguments `target` and `output` must have the same shape. "
            "Received: "
            f"target.shape={target.shape}, output.shape={output.shape}"
        )

    if from_logits:
        output = sigmoid(output)

    output = np.clip(output, backend.epsilon(), 1.0 - backend.epsilon())
    bce = target * np.log(output)
    bce += (1.0 - target) * np.log(1.0 - output)
    return -bce


def moments(x, axes, keepdims=False, synchronized=False):
    if synchronized:
        raise NotImplementedError(
            "Argument synchronized=True is not supported with NumPy."
        )
    axes = tuple(axes) if isinstance(axes, list) else axes
    # The dynamic range of float16 is too limited for statistics. As a
    # workaround, we simply perform the operations on float32 and convert back
    # to float16
    need_cast = False
    ori_dtype = backend.standardize_dtype(x.dtype)
    if ori_dtype == "float16":
        need_cast = True
        x = cast(x, "float32")

    mean = np.mean(x, axes, keepdims=True)

    # The variance is computed using $Var = E[|x|^2] - |E[x]|^2$, It is faster
    # but less numerically stable.
    variance = np.mean(np.square(x), axis=axes, keepdims=True) - np.square(mean)

View on GitHub (pinned to 7a34a03db6)

When it happens

Trigger: Thrown at keras/src/backend/numpy/nn.py:936 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/0a314c6c1d186cd1. Report an issue: GitHub.