INDEX
Explanations
attends to action tokens marked as "click" from context tokens marked with "on" or similar instructions
New Auto-Interp
Head Attr Weights
0:0.12
1:0.30
2:0.12
3:0.06
4:0.06
5:0.09
6:0.05
7:0.15
Negative Logits
Delete
-0.26
apost
-0.25
SpringRunner
-0.25
styleType
-0.24
Ανακτήθηκε
-0.24
tick
-0.24
atás
-0.24
delete
-0.23
Див
-0.23
SceneManagement
-0.23
POSITIVE LOGITS
AsUp
0.39
%)$
0.39
__':
0.36
abestanden
0.34
]--;
0.34
)";
0.33
%");
0.33
baku
0.33
EDEFAULT
0.33
"]));
0.31
Activations Density 0.062%