Taking different loss based on conditions in a batch

[ParGG (Gian Guido Parenza)]

Hi, I recently replied to a few similar topics. You can check them out, you should get some insight about how to solve your problem

Conditional loss for multi-class multi-output model

Hi Tal, when you load your data, on top of creating the one hot encoded classes, create an additional tensor that will mask your loss. If your room is empty, this tensor will be [1, 0, 0], whilst if it is not empty, it will be like [1, 1, 1]. When you compute your final loss (which is probably the sum of loss_r, loss_s, loss_b), just perform the internal (dot) product with the mask I told you about. In this case you will have the following situations: # empty room, mask = [1, 0, 0] loss = to…

Selection and application of available methods based on information from other tensor

I have a few approaches, but here are two. Let’s start with the one I like the most: Loss contribution masking: apply all the functions to the incoming input, and get the resulting vector [func1(input), func2(input), ..., func10(input)] and then perform the internal product (dot product) with the one hot encoded label vector. This will give you as loss just the value of the matching function. Example: class = 2 -> [0, 1, 0, ...], the loss will be loss = 0 * func1(input) + 1 * func2(input) + 0 …

If not, I can write something more personalised