INDEX

Explanations

use of

The neuron strongly activates on the function words “use” and “of,” i.e. common method‐section connectors in formal text.

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

for

-1.59

on

-1.57

to

-1.49

if

-1.43

 effect

-1.30

𝑎

-1.27

 since

-1.25

not

-1.24

 before

-1.23

so

-1.23

POSITIVE LOGITS

 نداره

1.71

了你

1.65

 véritable

1.55

 Además

1.54

新的

1.52

 beschränkt

1.52

 erklär

1.52

 dwie

1.51

 étranger

1.51

 ermöglich

1.49

Activations Density 0.063%