MCPcopy Create free account
hub / github.com/MARIO-Math-Reasoning/Super_MARIO / puct

Method puct

offline_inference.py:48–51  ·  view source on GitHub ↗
(self)

Source from the content-addressed store, hash-verified

46 prune: bool = False
47
48 def puct(self) -> float:
49 q_value = self.q_value if self.visit_count > 0 else 0
50 u_value = self.c_puct * self.prior * np.sqrt(self.parent.visit_count) / (1 + self.visit_count)
51 return q_value + u_value
52
53
54def rebuild_tree(

Callers 1

sort_by_strategyFunction · 0.45

Calls

no outgoing calls

Tested by

no test coverage detected