INDEX

Explanations

used as or for

The neuron primarily detects the word “used.”

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

 budow

-0.93

łąd

-0.90

hoku

-0.89

 nieco

-0.88

の中から

-0.87

varrho

-0.86

ocal

-0.85

шы

-0.85

 procesos

-0.85

nå

-0.84

POSITIVE LOGITS

for

1.72

 with

1.59

on

1.37

 indoors

1.35

at

1.30

as

1.28

 anywhere

1.25

by

1.22

in

1.14

 together

1.10

Activations Density 0.008%