Unverified50% confidenceFactExact time
Recurrent history can actively expand the available reward scale.
1
Sources
50%
Confidence
Long-term
Relevance
9/30/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe NGU algorithm superimposes external rewards with two types of intrinsic rewards: a short-term reward based on intra-episode novelty and a long-term reward based on lifetime novelty.71% similarUnverified贝尔曼方程将长期回报分解为即时奖励加上未来折扣回报的递归关系69% similarUnverifiedReward engineering is essentially a highly iterative process: design the reward, observe behavior, discover loopholes, correct the reward, and repeat.67% similarUnverified在智能体工作流中，历史交互作为输入Token累积，导致总Token消耗随任务轮次呈近似平方级增长66% similarUnverifiedCheckpoint partial rewards densify sparse signals by setting milestone-style rewards at intermediate nodes in long tasks, helping gradient propagation.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/961087API
curl https://kongchang.com/api/v1/knowledge/claims/961087MCP
get_claim(id=961087)