Dense Reward
A reward setup in reinforcement learning that provides a measure of progress at each step or milestone, yielding a smooth learning gradient, provided that an objective definition of progress exists.
Timeline (last 90 days)
Dense rewards can provide a smooth learning gradient, but only on the premise that an objective definition of "progress" must exist; otherwise every step's feedback is a subjective judgment, non-reproducible and impossible to scale.
If the specification is not written specifically enough, dense rewards will be confidently wrong and cause the Agent to optimize directly into the gap between the specification and reality.
All Facts (2)
Dense rewards can provide a smooth learning gradient, but only on the premise that an objective definition of "progress" must exist; otherwise every step's feedback is a subjective judgment, non-reproducible and impossible to scale.
50%UnverifiedIf the specification is not written specifically enough, dense rewards will be confidently wrong and cause the Agent to optimize directly into the gap between the specification and reality.
50%