AIF-C01 catalog scenario Modern Learning Paradigms Which description best defines self-supervised learning? A The model creates proxy labels from unlabeled data, such as predicting masked tokens. B The model clusters records only by distance and never optimizes a predictive objective. C The model learns only from human-labeled input-output pairs for the target task. D The model chooses actions in an environment and improves from delayed rewards. Answer ready for review Self-supervised learning is valuable when unlabeled data is abundant because the model can learn useful representations before supervised fine-tuning. Submit Answer Previous Start Quiz Session Open Explanation