Unverified50% confidenceFactExact time
后训练的最优解是参考分布的指数倾斜，即最优分布密度正比于参考模型密度乘以奖励的指数：π*(x) ∝ π_ref(x) · exp(r(x)/β)
1
Sources
50%
Confidence
Long-term
Relevance
7/15/2026
First Seen
Sources
Related Claims
UnverifiedThe posterior distribution is obtained via Bayes' theorem: p(θ|x) ∝ p(x|θ)·p(θ), combining prior knowledge with the data likelihood.71% similarUnverified重参数化技巧将采样表示为 z = μ + σ × ε（ε从标准正态分布采样），使随机性外置于ε，从而让μ和σ的梯度可正常反向传播实现端到端训练68% similarUnverifiedDuring training, LLMs optimize parameters by maximizing the prediction probability of the next token, meaning the model learns the statistical distribution of the training corpus rather than human aesthetic judgment.66% similarVerifiedDistribution shift refers to the systematic difference between the statistical distribution of data seen during model training and the distribution of data in the actual deployment environment.66% similarUnverifiedLLM天然倾向于输出最符合训练数据分布的平均化表达，这是AI内容高度同质化的根本原因66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/518181API
curl https://kongchang.com/api/v1/knowledge/claims/518181MCP
get_claim(id=518181)