INDEX

Explanations

the followed by circuit or ordinal

The neuron responds to occurrences of the definite article “the.”

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

 usamos

-2.42

嘥

-2.17

皤

-2.16

 noiva

-2.14

 gafas

-2.13

 boneca

-2.09

 первая

-2.09

 ruinas

-2.09

 dedicado

-2.08

POSITIVE LOGITS

</h3>

3.06

Of

2.63

The

2.61

 awhile

2.39

 example

2.39

</sub>

2.34

 Erwachsene

2.34

2.33

 Jugendliche

2.33

for

2.27

Activations Density 0.098%