Loss function for multi-binary output

I have supervised model that has 2 outputs- 1 is a float and the other is multi-binary (an array of 0/1). I have a loss function for each of the outputs and during training I see that the loss for the float is decreasing as it should but the loss of the multi-binary output stays high. The loss function I use for it is binary cross entropy, does anyone has a suggestion for other loss functions that can fit here?