Translations:Softmax Function/10/zh: Difference between revisions

Revision as of 22:04, 27 April 2026

Information about message (contribute)

This message has no documentation. If you know where or how this message is used, you can help other translators by adding documentation to this message.

Message definition (Softmax Function)

As the gap between the largest {{Term|logits|logit}} and the others increases, the output approaches a {{Term|one-hot encoding|one-hot}} vector. This "winner-take-most" behavior makes softmax well-suited for classification where a single class should dominate.

随着最大 logit 与其他 logit 之间的差距增大，输出会趋近于 one-hot 向量。这种"赢者通吃"的行为使 softmax 非常适合单一类别应占主导地位的分类任务。

Revision as of 03:17, 27 April 2026 (view source) DeployBot (talk \| contribs) (Batch translate Softmax Function unit 10 → zh) Tag: translation		Revision as of 22:04, 27 April 2026 (view source) DeployBot (talk \| contribs) (Batch translate Softmax Function unit 10 → zh) Tag: translation Newer edit →
Line 1:		Line 1:
	随着最大 logit 与其他 ~~logits 之间的差距增大，输出会趋近于一个~~ one-hot 向量。这种"赢者通吃"的行为使 softmax 非常适合单一类别应占主导地位的分类任务。		随着最大 logit 与其他 logit 之间的差距增大，输出会趋近于 one-hot 向量。这种"赢者通吃"的行为使 softmax 非常适合单一类别应占主导地位的分类任务。