INDEX
Explanations
speaker and title attribution
New Auto-Interp
Negative Logits
)$.
0.32
natürlich
0.30
—
0.30
💻
0.30
۔
0.30
多分
0.30
Possibly
0.29
በተጨማሪ
0.29
🇿
0.29
Polynomial
0.29
POSITIVE LOGITS
who
0.60
spokesman
0.59
kteří
0.52
spokesperson
0.51
who
0.50
president
0.49
spokes
0.49
którzy
0.49
spokeswoman
0.48
ktorí
0.48
Activations Density 0.001%