policies.py 文件源码

def _normal(self, x, mean, logstd):
        std = logstd.exp()
        std_sq = std.pow(2)
        a = (-(x - mean).pow(2) / (2 * std_sq)).exp()
        b = (2 * std_sq * self.pi.expand_as(std_sq)).sqrt()
        return a / b