More doubts on masking - Runnable example

[ordabayev]

@artistworking

c = sample("c", dist.Categorical(logits=torch.Tensor([[3, 5], [10, 8]])).to_event(1),infer={'enumerate': 'parallel'})

If you use .to_event then the distribution cannot be enumerated. Here is the issue I opened about it a while ago.