Concept
RL
RL stands for Reinforcement Learning, a training paradigm where a model learns by acting in an environment and receiving reward signals, rather than by fitting labeled input-output pairs. If supervised learning is "here is the correct answer, match it," RL is "here is a…
The rest of “RL” is a premium feature: every concept in the library gets a precise, practitioner-focused write-up like this one, cross-linked straight from the paper summaries that use it.
Log in to unlock→