Can loss be negative?
Some objectives can produce negative numerical values, depending on their definition. Common squared-error and standard cross-entropy losses are nonnegative under their usual assumptions.
Explore This Question
Search for additional explanations using these tools:
More FAQs in How does a loss function measure the quality of a model’s prediction?
No. Accuracy counts correct decisions, while a loss can measure confidence and error size. Two models with equal accuracy can have different losses.
Validation loss measures fit on examples excluded from training updates. A growing gap between training and validation performance can indicate overfitting or distribution differences.
Still have questions?
Browse all our FAQs or reach out to our support team