INDEX
Explanations
words related to heat or excitement
New Auto-Interp
Neuron Alignment
Index
Value
% of L₁
1387
+0.12
0.5%
1350
+0.11
0.4%
555
+0.10
0.4%
Correlated Neurons
Index
P. Corr.
Cos Sim.
1387
+0.12
0.02
257
+0.11
0.02
404
+0.10
0.02
Negative Logits
Berries
-0.52
liquido
-0.48
Pesto
-0.47
Cheesecake
-0.45
Cupcake
-0.44
Crème
-0.44
berge
-0.43
glPushMatrix
-0.43
Gateway
-0.43
Mousse
-0.42
POSITIVE LOGITS
Hot
1.24
Hot
1.24
hot
1.22
HOT
1.20
hot
1.14
HOT
1.11
hotter
0.93
hottest
0.81
saurait
0.80
hotspots
0.78
Activations Density 0.050%