INDEX
Explanations
linking phrases to specific requests
The neuron detects numeric tokens—multi-digit numbers and numerical values (including decimals and large numerals).
New Auto-Interp
Negative Logits
Its
0.24
its
0.23
servers
0.21
nodes
0.21
server
0.21
itself
0.20
metadata
0.20
engines
0.19
multil
0.19
stored
0.19
POSITIVE LOGITS
Concluding
0.21
uestas
0.19
тических
0.18
additional
0.18
ubah
0.18
комплекс
0.18
allaitement
0.18
Об
0.18
кы
0.18
Международ
0.18
Activations Density 0.263%