INDEX
Explanations
Romanian text discussing medical conditions and treatments
New Auto-Interp
Neuron Alignment
Index
Value
% of L₁
1839
+0.15
0.5%
1101
+0.11
0.3%
1387
+0.10
0.3%
Correlated Neurons
Index
P. Corr.
Cos Sim.
188
+0.15
0.03
282
+0.11
0.03
1839
+0.10
0.03
Negative Logits
împre
-1.19
munc
-1.03
piese
-0.98
toti
-0.98
oamen
-0.97
câte
-0.93
toată
-0.92
ziua
-0.91
argint
-0.91
oamenii
-0.90
POSITIVE LOGITS
Romanian
0.94
Romania
0.86
Bucharest
0.81
berea
0.79
inappro
0.76
mientras
0.73
vectorial
0.73
Diction
0.72
Lucian
0.72
impractica
0.72
Activations Density 0.140%