INDEX
Explanations
contrasting conjunctions and nouns
New Auto-Interp
Negative Logits
rather
0.46
piuttosto
0.45
rather
0.42
terbesar
0.40
இருந்தாலும்
0.40
തന്നെയാണ്
0.40
наверное
0.40
,|\
0.39
よりも
0.38
Rather
0.37
POSITIVE LOGITS
hingegen
1.62
dagegen
1.33
natomiast
1.11
viszont
1.05
మాత్రం
1.01
चाहिँ
0.99
却
0.96
卻
0.95
revanche
0.93
Conversely
0.89
Activations Density 0.026%