MeanAbsolutePercentageError

Computes the mean absolute percentage error between y_true & y_pred.

Abstract Signature:

MeanAbsolutePercentageError(y_true: Tensor, y_pred: Tensor)

PyTorch

API: —
Strategy: Macro 'torch.mean(torch.abs(({y_true} - {y_pred}) / torch.clamp({y_true}.abs(), min=1e-7))) * 100'

JAX (Core)

API: optax.mape_loss
Strategy: Direct Mapping
Official Docs ↗

Keras

API: keras.losses.mean_absolute_percentage_error
Strategy: Direct Mapping
Official Docs ↗

TensorFlow

API: tf.keras.losses.mean_absolute_percentage_error
Strategy: Direct Mapping
Official Docs ↗

Flax NNX

API: optax.mape_loss
Strategy: Direct Mapping
Official Docs ↗

PaxML / Praxis

API: optax.mape_loss
Strategy: Direct Mapping
Official Docs ↗