INDEX
Explanations
common phrases in financial and legal texts
New Auto-Interp
Neuron Alignment
Index
Value
% of L₁
184
+0.54
2.1%
50
+0.17
0.7%
1150
+0.15
0.6%
Correlated Neurons
Index
P. Corr.
Cos Sim.
184
+0.54
0.02
16
+0.17
0.03
50
+0.15
0.02
Negative Logits
jgl
-0.63
."\
-0.59
<bos>
-0.56
;++
-0.56
('\\-0.53
Italijani
-0.51
.$_
-0.51
AssemblyCulture
-0.51
."<
-0.50
Rumuni
-0.49
POSITIVE LOGITS
thut
1.04
Juf
0.97
impractica
0.95
aen
0.94
fta
0.94
fuf
0.92
affor
0.90
fte
0.89
ftu
0.88
guarante
0.87
Activations Density 0.039%