Can two correct answers have different probabilities?
Yes. Their phrasing, frequency in training patterns, and compatibility with the prompt can differ. Probability under the model is not the same as correctness.
Explore This Question
Search for additional explanations using these tools:
More FAQs in How does an LLM assign probabilities to possible next tokens?
A logit is an unnormalized model score. Softmax converts a collection of logits into nonnegative probabilities that sum to one across the available vocabulary.
No. Each selected token changes the sequence used for the next prediction. The model therefore computes a new conditional distribution as generation progresses.
Still have questions?
Browse all our FAQs or reach out to our support team