Ground Truth
Last reviewed by Moderation API
Ground truth is the human-labeled reference set that a classifier is trained and evaluated against, representing the correct answer for each example. The quality of the ground truth sets a hard ceiling on how good any model trained or measured against it can be.
