Unverified50% confidenceFactExact time
Contextual bandit is a class of sequential decision-making frameworks in reinforcement learning: the agent observes contextual features each round, selects an action and receives only an immediate reward, without involving the step-by-step credit assignment problem and without requiring temporal backup estimation of a value function.
1
Sources
50%
Confidence
Long-term
Relevance
10/30/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe multi-armed bandit is the simplest decision model in reinforcement learning, in which an agent chooses one of a fixed K actions to obtain an immediate reward, with no state transitions.71% similarUnverifiedIn the context of reinforcement learning, a rollout refers to the process of an agent completing one full pass through an environment or policy to generate a single trajectory.65% similarUnverifiedThe supervisor is a state machine that runs in DUNGEON mode most of the time with decisions made by the reinforcement learning model, and switches to PATHFIND mode after the step budget is exhausted or enough monsters are killed, using BFS to pathfind to the most recently seen stairs.65% similarUnverifiedIf a system prompt contains a meta-instruction allowing user instructions to override safety training, it is equivalent to pre-installing a priority-inversion switch in the model's decision logic, enabling attackers to bypass safeguards without breaking the model itself.64% similarUnverifiedThe core of gamified learning lies in breaking down knowledge systems into digestible small units through mechanisms such as segmented tasks, instant feedback, and progress incentives.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1114434API
curl https://kongchang.com/api/v1/knowledge/claims/1114434MCP
get_claim(id=1114434)