Dataset and label checks · Question 39 of 60
What can a data point with high loss during model training indicate?
1 answer is correct. Medium
Show answer
Answer
- A A possibly wrong label, because the model struggles to learn the assigned label
Explanation
In model loss analysis, data points with high loss show that the predictions deviate substantially from their labels. This can indicate mislabeling, because the model struggles to learn the assigned label.