All translations

Enter a message name below to show all available translations.

Message

Found 3 translations.

NameCurrent message text
 h English (en)where <math>m = \max_j z_j</math>. Subtracting the maximum {{Term|logits|logit}} ensures the largest exponent is zero, preventing overflow. All major {{Term|deep learning}} frameworks implement this fused operation (e.g., PyTorch's <code>CrossEntropyLoss</code> accepts raw {{Term|logits}}).
 h Spanish (es)donde <math>m = \max_j z_j</math>. Restar el {{Term|logits|logit}} máximo asegura que el mayor exponente sea cero, evitando el desbordamiento. Todos los frameworks principales de {{Term|deep learning|aprendizaje profundo}} implementan esta operación fusionada (p. ej., <code>CrossEntropyLoss</code> de PyTorch acepta {{Term|logits}} sin procesar).
 h Chinese (zh)其中<math>m = \max_j z_j</math>。减去最大{{Term|logits|logit}}可确保最大指数为零,从而防止上溢。所有主流的{{Term|deep learning|深度学习}}框架都实现了这一融合操作(例如,PyTorch的<code>CrossEntropyLoss</code>接受原始{{Term|logits}})。