ranked.py 文件源码

python
阅读 23 收藏 0 点赞 0 评论 0

项目:openai_lab 作者: kengz 项目源码 文件源码
def __init__(self, env_spec, **kwargs):
        super(HighLowMemory, self).__init__(env_spec)
        # use the old self.exp as buffer, remember to clear
        self.last_exp = self.exp
        self.epi_memory_high = []
        self.epi_memory_low = []
        self.max_reward = -math.inf
        self.min_reward = math.inf
        # 1st  5 epis goes into bad half, recompute every 5 epis
        self.threshold = math.inf
        self.threshold_history = []
        self.epi_num = 0
        self.prob_high = 0.66
        self.num_epis_to_sample = 3
        self.max_epis_in_mem = 15
        self.recompute_freq = 10
        log_self(self)
评论列表
文章目录


问题


面经


文章

微信
公众号

扫码关注公众号