Policy Gradients: Learning to Aim

๐ŸŽฏ Policy Gradients

Try it. Score it. Nudge the odds. Repeat.

0Throws
โ€”Avg score
WideAim spread

That's the whole algorithm: an action that beats its own average gets more likely, one that falls short gets less likely. No one tells it the right answer โ€” the odds just drift toward what works.