≥

The passage around this formula

Reinforcement learning’s standard tool for turning a sequence of actions and a single delayed reward into a training signal is the policy gradient, and its textbook reward-to-go form is $\nabla_\theta J(\theta) = \mathbb{E}_\pi\!\left[\sum_{t=0}^{T} \nabla_\theta \log \pi_\theta(a_t \mid s_t) \left(\sum_{t' \ge t} r_{t'}\right)\right]$. Read the inner sum literally: every action in the trajectory is credited with everything that happens from that point on, not with its own specific contribution. When T is small and rewards are dense, that crude attribution washes out quickly. When T is large and the reward is a single terminal signal — a multi-step coding task that either compiles and passes its tests or does not, a multi-turn support conversation that either resolves the case or does not — the same sum assigns identical…

Read this part in the article →