INDEX
Explanations
references to legal cases and procedures
New Auto-Interp
Neuron Alignment
Index
Value
% of L₁
1609
+0.09
0.2%
1473
+0.08
0.2%
581
+0.08
0.2%
Correlated Neurons
Index
P. Corr.
Cos Sim.
1036
+0.09
0.01
488
+0.08
0.05
1782
+0.08
0.04
Negative Logits
meis
-0.95
wien
-0.89
shutterstock
-0.88
stockholm
-0.88
italia
-0.86
reluct
-0.85
elek
-0.84
pixabay
-0.83
bordeaux
-0.83
hek
-0.82
POSITIVE LOGITS
IsContent
0.58
<bos>
0.53
fjspx
0.53
Another
0.52
another
0.52
ワイイ
0.51
CodedInputStream
0.49
RemoveField
0.48
ViewById
0.48
Another
0.48
Activations Density 0.446%