Unverified50% confidenceFactExact time
On-policy RL requires the agent to continuously interact with the environment during training, and the interaction cost grows linearly with training scale.
1
Sources
50%
Confidence
Long-term
Relevance
10/21/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedBehavioral cloning suffers from distribution shift; improved algorithms such as DAgger address this by having the learning agent continuously interact with the expert to expand the training data.75% similarUnverified实时在线学习要求智能体在连续流动的时间中同步完成感知、决策与参数更新，与大语言模型训练-推理分离的范式形成对比74% similarUnverifiedOnline reinforcement learning faces the exploration-exploitation trade-off, where an agent must balance exploiting known effective strategies against trying new behaviors.73% similarUnverifiedIn the Agent context, reinforcement learning (RL) is mainly used to optimize decision-making policies, with commonly used methods including RLHF and DPO; compared with supervised fine-tuning, it can explore better paths through trial and error, but it faces challenges such as difficulty in reward design, low sample efficiency, and unstable training.73% similarUnverifiedIncremental Learning or Online Learning refers to a model continuously updating its own parameters as new data is fed in, rather than completing training in a single pass.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1079158API
curl https://kongchang.com/api/v1/knowledge/claims/1079158MCP
get_claim(id=1079158)