INDEX
Explanations
phrases containing symbols (e.g., âĢ, ľ) commonly used for emphasis or decoration
the presence of end-of-text tokens indicating the completion of sections or ideas
New Auto-Interp
Negative Logits
gad
-0.72
scattering
-0.71
dispers
-0.70
anwhile
-0.69
ierrez
-0.68
casting
-0.68
scatter
-0.66
nearest
-0.66
detached
-0.65
Peb
-0.64
POSITIVE LOGITS
º
1.23
¹
1.11
£
1.08
®
1.05
į
1.04
ı
1.02
Į
1.02
Ī
1.00
¦
0.98
¬
0.96
Activations Density 0.107%