Closed xxxyyycc closed 4 years ago
def cost_fn(logits, levels, imp): val = (-torch.sum((F.logsigmoid(logits)*levels + (F.logsigmoid(logits) - logits)*(1-levels))*imp, dim=1)) return torch.mean(val)
The second term of the cost seems a little different from the original definition,could you explain it? thanks.
The second term of the cost seems a little different from the original definition,could you explain it? thanks.