Pretty sure there’s barely any meaningful Training data for this? Would be interesting to see these models return a kind of coefficient indicating how “often”

Pretty sure there’s barely any meaningful Training data for this? Would be interesting to see these models return a kind of coefficient indicating how “often” a semantic context has appeared in the training data..