INDEX
Explanations
performance comparison and results
New Auto-Interp
Negative Logits
финансо
0.64
कानून
0.61
电子商务
0.61
політи
0.55
电商
0.55
宪
0.55
полити
0.54
официа
0.54
公开
0.54
유명
0.54
POSITIVE LOGITS
consistently
0.63
results
0.59
decreased
0.55
low
0.54
moderately
0.54
scores
0.53
slightly
0.52
almost
0.52
showed
0.51
outperforms
0.51
Activations Density 0.036%