Thank you very much for sharing Aligned MTL. After comparing your experiments with the paper, I have a question regarding the pseudocode in the first image. In the pseudocode, 'w' represents the weight for each loss function. However, in your code, you directly sum over 'B' in the 'alpha' code. I was wondering if you assume that the weights for each task are equal when writing the code?
Thank you very much for sharing Aligned MTL. After comparing your experiments with the paper, I have a question regarding the pseudocode in the first image. In the pseudocode, 'w' represents the weight for each loss function. However, in your code, you directly sum over 'B' in the 'alpha' code. I was wondering if you assume that the weights for each task are equal when writing the code?
