Marginal probabilities from a CRF

Hi! I was wondering whether there’s an easy way (i.e., there’s code available) to get the marginal probabilities for a particular label at a particular timestep from the AllenNLP CRF. In particular, I’d like a probability distribution over my labels at each timestep.

1 Like

I don’t think we have any code available for this. If anyone else does, feel free to speak up! Contributions welcome.