INDEX
Explanations
words related to problems, challenges, and exacerbation in social contexts
New Auto-Interp
Neuron Alignment
Index
Value
% of L₁
889
+0.14
0.5%
1616
+0.13
0.4%
1872
+0.12
0.4%
Correlated Neurons
Index
P. Corr.
Cos Sim.
1872
+0.14
0.03
890
+0.13
0.03
889
+0.12
0.02
Negative Logits
abbra
-0.52
résult
-0.50
FetchType
-0.50
rimb
-0.47
Personne
-0.47
dirit
-0.47
Avez
-0.47
ioutil
-0.47
équi
-0.46
alis
-0.46
POSITIVE LOGITS
enhanced
0.71
Enhanced
0.64
enhanced
0.64
enhance
0.63
enhancing
0.62
enhancement
0.60
enhancements
0.60
zove
0.60
pymysql
0.58
escalated
0.57
Activations Density 0.140%