如何在张量流对象检测中使用指数学习率？

Question

谁能告诉如何在配置文件中设置 exponential learning_rate 而不是恒定学习率？

配置文件中的恒定学习率：

learning_rate: {
            manual_step_learning_rate {
              initial_learning_rate: 0.003
              schedule {
                step: 6000
                learning_rate: .0003
              }
              schedule {
                step: 12000
                learning_rate: .00003
              }

Answer 1

参考 https://github.com/tensorflow/models/blob/master/research/object_detection/protos/optimizer.proto

例子：

learning_rate: {
        exponential_decay_learning_rate {
          initial_learning_rate: 0.004
          decay_steps: 800720
          decay_factor: 0.95
        }
      }