INDEX

Explanations

prepositions and conjunctions

The neuron fires on rare or infrequently used tokens—often named entities or specialized/domain-specific words.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 altres

-1.04

گوشی

-1.04

gleiche

-1.04

 tää

-1.02

giendo

-1.02

是什麼

-1.02

țiile

-1.02

corrhi

-1.00

為に

-1.00

ėjimas

-1.00

POSITIVE LOGITS

in

2.09

to

1.83

 with

1.59

 into

1.50

by

1.50

 from

1.48

 به

1.45

on

1.41

for

1.36

 ability

1.34

Activations Density 0.000%