INDEX

Explanations

frequently appears

The neuron consistently lights up on descriptive content words—adjectives, adverbs, and other qualifiers that add qualitative detail or emphasis.

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

了一些

-1.33

 juſ

-1.31

 normaux

-1.24

 inoltre

-1.23

 grotte

-1.22

ួ

-1.21

DALE

-1.16

垪

-1.15

 paroi

-1.14

 verschillende

-1.13

POSITIVE LOGITS

Both

1.38

and

1.38

 every

1.34

any

1.34

 both

1.27

After

1.18

all

1.16

its

1.08

 epitom

1.04

Usually

1.03

Activations Density 0.028%