INDEX
Explanations
elements related to comedy shows or stand-up performances
New Auto-Interp
Neuron Alignment
Index
Value
% of L₁
946
+0.18
0.6%
453
+0.12
0.4%
1385
+0.11
0.3%
Correlated Neurons
Index
P. Corr.
Cos Sim.
946
+0.18
0.04
453
+0.12
0.05
378
+0.11
0.02
Negative Logits
reluct
-1.54
accla
-1.49
impra
-1.49
guarante
-1.47
increa
-1.46
depic
-1.44
snoopy
-1.44
disagre
-1.43
shenan
-1.42
purcha
-1.42
POSITIVE LOGITS
<bos>
0.93
GRAPHY
0.62
ngdoc
0.58
person
0.55
dish
0.54
album
0.54
module
0.54
item
0.54
object
0.53
article
0.53
Activations Density 0.333%