INDEX

Explanations

equal to the number of

The neuron never activates—it appears to be effectively “dead” and does not respond to any token.

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

 Ense

-1.41

and

-1.38

THERE

-1.37

 renom

-1.30

 velours

-1.29

桠

-1.29

 Jogo

-1.27

 warum

-1.25

ครั้ง

-1.24

 cortina

-1.24

POSITIVE LOGITS

 that

2.27

its

1.89

 your

1.84

the

1.60

you

1.59

 those

1.51

it

1.45

if

1.43

 españolas

1.38

 gdyby

1.36

Activations Density 0.113%