在对卷积层及池化层进行权重初始化时, 激活函数为 sigmoid 时: def weight_variable(shape): inital = tf.truncated_normal(shape) return tf.Variable(inital) 权重初始化为截断正态分布,默认参数如下: def truncated_normal(shape, mean=0.0, stddev=1.0, dtype=dtypes.float32,