Open mglowacki100 opened 9 months ago
I've looked at the function:
def score_dual(X, q=1, normalize=True): if normalize: X = preprocessing.normalize(X, axis=1) n = X.shape[0] S = X.T @ X w = scipy.linalg.eigvalsh(S / n) m = w > 0 return np.exp(entropy_q(w, q=q))
from https://github.com/vertaix/Vendi-Score/blob/main/vendi_score/vendi.py, what is a purpose of line m = w > 0
m = w > 0
Does it make sense to use score_dual for calculation ensemble diversity ?
score_dual
I've looked at the function:
from https://github.com/vertaix/Vendi-Score/blob/main/vendi_score/vendi.py, what is a purpose of line
m = w > 0
Does it make sense to use
score_dual
for calculation ensemble diversity ?