rts/game_CF/model.py [79:98]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
        policy = self.softmax(self.linear_policy(output))
        value = self.linear_value(output)
        return dict(V=value, pi=policy)

    def test(self):
        x = torch.cuda.FloatTensor(max_s, max_s)
        x.fill_(0)
        for i in range(max_s):
            x[i, i] = 1

        res = self(self._var(x))
        # Check both policy and value function
        print(res["pi"].exp())
        print(res["V"])

# Format: key, [model, method]
# if method is None, fall back to default mapping from key to method
Models = {
    "actor_critic": [Model_ActorCritic, ActorCritic]
}
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



rts/game_TD/model.py [71:90]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
        policy = self.softmax(self.linear_policy(output))
        value = self.linear_value(output)
        return dict(V=value, pi=policy)

    def test(self):
        x = torch.cuda.FloatTensor(max_s, max_s)
        x.fill_(0)
        for i in range(max_s):
            x[i, i] = 1

        res = self(self._var(x))
        # Check both policy and value function
        print(res["pi"].exp())
        print(res["V"])

# Format: key, [model, method]
# if method is None, fall back to default mapping from key to method
Models = {
    "actor_critic": [Model_ActorCritic, ActorCritic]
}
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -