docs: say what the two lines are showing

The block sat between two paragraphs with no lead-in, so the claim it makes -- that RL interaction has the same shape as token generation -- was left for the reader to infer and appeared nowhere on the card. It now says so, in the wording the repository README already uses. The RL line also pairs state with action and marks the next state as coming from the environment, which the old one-per-arrow reading blurred.

README.md

@@ -15,9 +15,11 @@ hybrid action spaces**, including cooperative and competitive multi-agent scenes
 without a separate model per action-space type. Autoregressive generation carries
 long-horizon control without auxiliary networks.
 
+Both LLM generation and RL interaction are autoregressive:
+
 ```text
-action → next state → next action      (RL rollouts)
-token  → next token  → next token      (LLM generation)
+token           → next token                           (LLM generation)
+(state, action) → (next state from env, next action)   (RL rollout)
 ```
 
 Stable under self-generated rollouts — long-horizon control without the drift that