INDEX
Explanations
instances of page references in documents
New Auto-Interp
Neuron Alignment
Index
Value
% of L₁
23
+0.20
1.1%
52
+0.13
0.7%
106
+0.11
0.6%
Correlated Neurons
Index
P. Corr.
Cos Sim.
23
+0.20
0.04
400
+0.13
0.03
77
+0.11
0.02
Negative Logits
wordpress
-1.74
jsfiddle
-1.73
iei
-1.72
fiddle
-1.71
cdn
-1.70
zent
-1.60
obox
-1.53
apis
-1.52
oxford
-1.52
wer
-1.49
POSITIVE LOGITS
sign
1.59
coming
1.56
move
1.55
*~(
1.48
cos
1.46
condensate
1.45
enter
1.45
stay
1.44
acquire
1.43
cdot
1.41
Activations Density 1.095%