Unverified50% confidenceFactExact time
Sampling regret is defined as the temperature-normalized cumulative KL divergence between the learner's distribution and the Gibbs target distribution, corresponding to the suboptimality gap in entropy-regularized reward maximization.
1
Sources
50%
Confidence
Long-term
Relevance
10/16/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedAn information-theoretic contrastive objective usually refers to a loss function that measures distributional differences such as KL divergence or cross-entropy, and NatScale uses the calibrated weights as the error penalty coefficients in these objectives.74% similarUnverified跨分词器蒸馏前应先测量熵保留率或概率质量对齐度以避免在错误目标上浪费算力70% similarVerifiedCross-entropy loss originates from KL divergence in information theory and measures the degree of difference between the true probability distribution and the predicted probability distribution.68% similarUnverified监督学习本质上是函数逼近问题，通过最小化损失函数实现，常见损失函数包括回归任务的均方误差（MSE）和分类任务的交叉熵（Cross Entropy）68% similarUnverified排查SAC训练崩溃应监控策略熵和α变化曲线，若熵持续下降至接近零说明探索消亡，可固定α为较大常数或重设目标熵为-action_dim68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1042175API
curl https://kongchang.com/api/v1/knowledge/claims/1042175MCP
get_claim(id=1042175)