INDEX

Explanations

prepositions, conjunctions and articles

The neuron fires on tokens that begin new sentences or clauses—i.e. it detects sentence-initial 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

 как

-1.34

 there

-1.34

^{}

-1.32

 есть

-1.32

-1.29

 graisse

-1.27

ladimir

-1.25

-1.24

ным

-1.22

 dAtA

-1.22

POSITIVE LOGITS

 when

1.66

 where

1.65

for

1.56

##

1.41

 saken

1.33

ági

1.32

 manipula

1.29

鮠

1.27

 dorp

1.26

 Vinci

1.23

Activations Density 0.098%