SimiaoZuo
about the calculation of intensity for different event types
In the calculation of f_k in the paper, each event type should correspond to a different beta_k. However, in the code, the model only has one beta parameter. Shouldn't there be a separate learnable beta_k parameter for each event type? Can someone please give me some guidance?
hello,I think this beta is a parameter in softplus: https://pytorch.org/docs/stable/generated/torch.nn.Softplus.html.
hello,I think this beta is a parameter in softplus: https://pytorch.org/docs/stable/generated/torch.nn.Softplus.html.
thank for your reply.I know that beta is a parameter in softplus .But my question is that Shouldn't there be a separate learnable beta_k parameter for each event type? It means if we have ten event types, we should have ten beta_k(0<k<=9).Can you please give me some guidance ?
I have the same question here, but I think they implemented the shared beta (k) instead, not beta_k. However, I have other question regarding this part. Where does the "current influence" from Eq.6 (paper page 7) be implemented in the code? From my understanding, when they calculate the type-specific intensity (\lambda_k) in the code (see pic I posted), only w^T_k*h(t_j) is calculated here, no?
