Unverified50% confidenceFactExact time
Common implementations of per-answer probability output include adding a classification head at the output layer, scoring candidate answers by likelihood, or normalizing candidates with a softmax over logits.
1
Sources
50%
Confidence
Long-term
Relevance
10/29/2026
First Seen
Sources
Related Entities
Related Claims
Unverified口语化采样通过在提示词中要求模型生成一组候选答案并显式说出每个答案对应的概率分布来提升多样性77% similarUnverifiedCLM computes probabilities over the candidate set passed in the request via dot product plus softmax, so its probabilities are relative to the provided candidate set and can only be used to rank candidates within the same request, not for absolute cross-request comparisons.74% similarUnverifiedFor multiple-choice questions, the logits over each option token can be normalized via softmax to obtain a probability distribution, where a highly concentrated distribution indicates the model is more confident in the answer73% similarUnverifiedPrompts act somewhat like a bias adjustment on the probability distribution: by providing context, roles or examples, they raise the probability of certain expression paths and lower others, thereby guiding the model toward more creative output.73% similarVerifiedLLMs generate output token by token in an autoregressive manner, with each word's selection based on probability sampling from the current context.73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1110152API
curl https://kongchang.com/api/v1/knowledge/claims/1110152MCP
get_claim(id=1110152)