https://github.com/vb000/LookOnceToHear/blob/main/src/models/tfgridnet_realtime/film.py#L21 Hi, I think for the FiLM Part, maybe the bias should be b = self.bias(embedding) instead of self.weights?