You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In figure 2B of the first moseq paper there is a graph for syllable cross-likelihood, would like to see that added as an additional verification that the model is not generating overlapping syllables. Thank you!
The text was updated successfully, but these errors were encountered:
Hi. I would upvote this - it is still very actual for comparing the models. By the way, I've found the function get_crosslikes that should do the trick in the moseq2_model.train.util but it seems to be a legacy code incompatible with current variables and stuff. If somebody could clear this out I can try to implement it myself. Thank you anyway.
I've tried using get_crosslikes with the saved model instances. It seems that the saved state objects, defined in pyhsmm, in model.states_list have their state.data attribute set to None. This attribute is required to be filled to instantiate other state attributes such as state.aBl. Not sure how to move forward from here as I'm not familiar with pyhsmm internals at all.
In figure 2B of the first moseq paper there is a graph for syllable cross-likelihood, would like to see that added as an additional verification that the model is not generating overlapping syllables. Thank you!
The text was updated successfully, but these errors were encountered: