INDEX
Explanations
attends to the punctuation mark "period" from later tokens that contain numerical values or data representations
New Auto-Interp
Head Attr Weights
0:0.09
1:0.24
2:0.09
3:0.09
4:0.08
5:0.14
6:0.10
7:0.14
Negative Logits
تضيفلها
-0.35
Geplaatst
-0.33
帖最后由
-0.31
Спасылкі
-0.29
esternos
-0.29
Q
-0.29
اریخ
-0.29
rispetto
-0.27
mun
-0.27
c
-0.26
POSITIVE LOGITS
)}</
0.49
CloseOperation
0.42
gebob
0.39
WindowConstants
0.38
}}"></
0.37
addPreferredGap
0.36
.)}
0.36
ьаж
0.35
[])
0.35
*********/
0.35
Activations Density 0.005%