INDEX
Explanations
the presence of specific numerical or relational concepts, particularly focusing on the word "to" and variations thereof
New Auto-Interp
Negative Logits
sợi
-0.50
-0.41
Yours
-0.41
2
-0.40
FieldNumber
-0.40
1
-0.38
mathrm
-0.37
verdad
-0.37
oprecip
-0.37
তথ্যসূত্র
-0.36
POSITIVE LOGITS
<bos>
1.16
AddTagHelper
0.89
0.85
esModule
0.84
'}>
0.84
"}>
0.83
)':
0.79
)_/¯
0.79
")->
0.78
"}";
0.77
Activations Density 0.693%