建立一個全連接層的神經網絡weight[784,10] bias[10] with tf.variable_scope("full_connected_model"): # 隨機初始化權重和偏置 weight = tf.
確定! 回上一頁