Unverified50% confidenceSolutionExact time
ORL adopts on-policy residual learning: training data is generated in real time by the current model to ensure that the training distribution matches the inference distribution.
1
Sources
50%
Confidence
Long-term
Relevance
10/17/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedFederated learning allows models to complete training updates locally on the data, uploading only gradients rather than raw data75% similarUnverifiedContinual training is a method of continuously training on newly added data to keep the model's knowledge up to date.75% similarVerifiedSelf-supervised learning is the core paradigm of the pre-training phase, where the model constructs training signals from the data itself through a next-token prediction task.74% similarUnverifiedIn traditional machine learning, HITL is commonly used in Active Learning, allowing the model to request human annotation on uncertain samples to improve training efficiency.74% 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73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1049945API
curl https://kongchang.com/api/v1/knowledge/claims/1049945MCP
get_claim(id=1049945)