Translations:Softmax Function/30/en: Difference between revisions
(Importing a new version from external source) |
(Importing a new version from external source) Tag: Manual revert |
||
| (2 intermediate revisions by the same user not shown) | |||
| Line 1: | Line 1: | ||
# A neural network produces raw logits <math>\mathbf{z}</math> from its final linear layer. | # A neural network produces raw {{Term|logits}} <math>\mathbf{z}</math> from its final linear layer. | ||
# Softmax converts logits to probabilities: <math>\hat{\mathbf{y}} = \sigma(\mathbf{z})</math>. | # Softmax converts {{Term|logits}} to probabilities: <math>\hat{\mathbf{y}} = \sigma(\mathbf{z})</math>. | ||
# The predicted class is <math>\hat{c} = \arg\max_k \hat{y}_k</math>. | # The predicted class is <math>\hat{c} = \arg\max_k \hat{y}_k</math>. | ||
# Training uses [[Cross-Entropy Loss]] applied to the predicted distribution and the true labels. | # Training uses [[Cross-Entropy Loss]] applied to the predicted distribution and the true labels. | ||
Latest revision as of 23:34, 27 April 2026
- A neural network produces raw logits $ \mathbf{z} $ from its final linear layer.
- Softmax converts logits to probabilities: $ \hat{\mathbf{y}} = \sigma(\mathbf{z}) $.
- The predicted class is $ \hat{c} = \arg\max_k \hat{y}_k $.
- Training uses Cross-Entropy Loss applied to the predicted distribution and the true labels.