TheAlgorithms/Python · error · ValueError

Shape of y_true and y_pred must be the same.

Error message

Shape of y_true and y_pred must be the same.

What it means

Thrown by categorical_focal_cross_entropy when the y_true (one-hot labels) and y_pred (predicted probabilities) matrices have different shapes. The loss is computed element-wise (alpha * (1 - y_pred)^gamma * y_true * log(y_pred)), so both inputs must have identical dimensions (samples x classes). Any shape divergence makes the element-wise product undefined, hence the early validation.

Source

Thrown at machine_learning/loss_functions.py:228

    ValueError: y_true must be one-hot encoded.

    >>> true_labels = np.array([[1, 0, 0], [0, 1, 0]])
    >>> pred_probs = np.array([[0.9, 0.1, 0.1], [0.2, 0.7, 0.1]])
    >>> categorical_focal_cross_entropy(true_labels, pred_probs)
    Traceback (most recent call last):
        ...
    ValueError: Predicted probabilities must sum to approximately 1.

    >>> true_labels = np.array([[1, 0, 0], [0, 1, 0], [0, 0, 1]])
    >>> pred_probs = np.array([[0.9, 0.1, 0.0], [0.2, 0.7, 0.1], [0.0, 0.1, 0.9]])
    >>> alpha = np.array([0.6, 0.2])
    >>> categorical_focal_cross_entropy(true_labels, pred_probs, alpha)
    Traceback (most recent call last):
        ...
    ValueError: Length of alpha must match the number of classes.
    """
    if y_true.shape != y_pred.shape:
        raise ValueError("Shape of y_true and y_pred must be the same.")

    if alpha is None:
        alpha = np.ones(y_true.shape[1])

    if np.any((y_true != 0) & (y_true != 1)) or np.any(y_true.sum(axis=1) != 1):
        raise ValueError("y_true must be one-hot encoded.")

    if len(alpha) != y_true.shape[1]:
        raise ValueError("Length of alpha must match the number of classes.")

    if not np.all(np.isclose(np.sum(y_pred, axis=1), 1, rtol=epsilon, atol=epsilon)):
        raise ValueError("Predicted probabilities must sum to approximately 1.")

    # Clip predicted probabilities to avoid log(0)
    y_pred = np.clip(y_pred, epsilon, 1 - epsilon)

    # Calculate loss for each class and sum across classes
    cfce_loss = -np.sum(

View on GitHub (pinned to f5988cc097)

Solutions

1. Print y_true.shape and y_pred.shape right before the call and make them equal, typically (num_samples, num_classes).
2. One-hot encode y_true with the same num_classes as y_pred's last axis (e.g. np.eye(num_classes)[y_true_int]).
3. If y_pred was transposed or reshaped during preprocessing, fix the reshaping so rows are samples and columns are classes.
4. Verify the model's final dense layer size matches the number of classes in the label encoder.

Example fix

# before
y_true = np.array([0, 2, 1])              # shape (3,)
loss = categorical_focal_cross_entropy(y_true, y_pred)  # y_pred shape (3, 3)

# after
y_true = np.eye(y_pred.shape[1])[y_true]   # shape (3, 3)
loss = categorical_focal_cross_entropy(y_true, y_pred)

if y_true.shape != y_pred.shape:
    raise ValueError(f"shape mismatch: {y_true.shape} vs {y_pred.shape}")
loss = categorical_focal_cross_entropy(y_true, y_pred, alpha)

def is_valid_cfce_input(y_true: np.ndarray, y_pred: np.ndarray) -> bool:
    return y_true.ndim == 2 and y_true.shape == y_pred.shape

try:
    loss = categorical_focal_cross_entropy(y_true, y_pred, alpha)
except ValueError as e:
    logger.error("focal loss input invalid: %s; shapes %s vs %s", e, y_true.shape, y_pred.shape)
    raise

When it happens

Trigger: Calling categorical_focal_cross_entropy(y_true, y_pred) where y_true.shape != y_pred.shape, e.g. labels one-hot encoded over 3 classes but predictions over 4 classes, or a 1-D label array paired with a 2-D probability matrix.

Common situations: Mismatch between the number of output units of the model and the one-hot encoder; forgetting to one-hot encode y_true (passing integer class labels of shape (N,) against y_pred of shape (N, C)); transposed predictions; batch slicing that drops a dimension.

Related errors

• Input data set must be one-dimensional
• Data set labels must be one-dimensional
• Input arrays must have the same shape.
• Length of predicted and actual array must be same.
• Data must have dimensions N x 1

AI-assisted analysis of TheAlgorithms/Python@f5988cc097 (2026-08-14). Data as JSON: /api/errors/0a8f9344b190fb49. Report an issue: GitHub.