Unverified50% confidenceFactExact time
SAC（Soft Actor-Critic）的核心机制是最大熵强化学习，在优化奖励的同时鼓励策略保持随机性，由温度系数α控制探索与利用的平衡
1
Sources
50%
Confidence
Long-term
Relevance
10/6/2026
First Seen
Sources
SAC训练崩溃之谜：策略熵坍塌与DIAYN调试指南
Related Entities
Cite This Claim
Stable URI
https://kongchang.com/claim/982248API
curl https://kongchang.com/api/v1/knowledge/claims/982248MCP
get_claim(id=982248)