How to create a PyCaffe loss layer with two tops?

[Raúl Gombru]

Hello,

I want to create a PyCaffe loss layer with two tops. The reason is that I want to output from the loss layer the loss and another value to monitor it. 

If I just add two tops as:

PyCaffe layer: 

top[0].data[...] = loss
top[1].data[...] = value_to_monitor

prototxt:

layer {
  name: "lossCustom"
  type: "Python"
  bottom: "fc2"
  bottom: "label"
  top: "loss"
  top: "value_to_monitor"
  python_param {
    module: "my_loss_layer"
    layer: "myLossLayer"
  }
  loss_weight: 1
}

I get this error:

Check failed: top.size() == num_loss_weights (2 vs. 1) loss_weight must be unspecified or specified once per top blob.

But I don't need to weight two losses (btw, how weight values should be written in the prototxt to do that?), I just need to output another value.

How can I do that?

Thanks

[Przemek D]

If you specify a loss_weight for a layer, you must do so for each top blob of that layer individually. If you have 2 top blobs, there must be 2 loss_weights. If you want to only use one top blob as a loss, I think it is enough to set a positive loss weight for that blob, while setting the one corresponding to your "monitor" blob to zero.
Try this:
layer {
  name: "myLayer"
  //...
  top: "loss"
  top: "monitor"
  //...
  loss_weight: 1
  loss_weight: 0
}

[Raúl Gombru]

Thanks! I understood that, but I didn't know the syntax. Closing.