Unverified50% confidenceFactExact time
当动作幅度趋近于零时,不同技能的状态轨迹难以区分,判别器梯度消失,形成恶性循环导致技能学习崩溃
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/982253API
curl https://kongchang.com/api/v1/knowledge/claims/982253MCP
get_claim(id=982253)