snorkel-team / snorkel

A system for quickly generating training data with weak supervision
https://snorkel.org
Apache License 2.0
5.81k stars 857 forks source link

getting negative value after label_model.get_conditional_probs() and snorkel LF #1733

Closed ShibaPrasad closed 5 months ago

ShibaPrasad commented 7 months ago

Hello All,

I am getting -ve values in the " conditional probabilities table" - after applying label_model.get_conditional_probs(). (snorkel LF) I am not sure why it will be -ve. (though my end-to-end flow working fine)

[[[-0.0461843 0.2952323 ] [ 0.99900001 0.67258263] [ 0.04718433 0.03218511]]

[[-0.00505793 0.29819989] [ 0.99900001 0.69790399] [ 0.00605786 0.0038961 ]]

[[-0.00505793 0.29819989] [ 0.99900001 0.69790399] [ 0.00605786 0.0038961 ]]

[[-0.00505793 0.29819989] [ 0.99900001 0.69790399] [ 0.00605786 0.0038961 ]]] (4, 3, 2)

ShibaPrasad commented 5 months ago

Solved