回归模型输出所有值的平均值

Question

我跟着Tensorflow的regression tutorial。我遇到的问题是，我的模型仅输出给定的每个输入的所有标签的平均值。我的模型将图像作为输入，并假定输出0.5-20之间的数字，但将所有图像标记为平均值12.913456。有任何想法吗？我的代码如下：

[[12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [28.889048]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [10.809496]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]
 [12.913456]]

我的预测向量：

{{1}}