train.AdamOptimizer, tf.train.FtrlOptimizer,... will give you better results than gradient descent.
確定! 回上一頁