INDEX

Explanations

in, into, and movement

The neuron is detecting numeric tokens (numbers, especially year‐like numbers).

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

if

-2.06

the

-2.02

your

-1.90

our

-1.84

 setelah

-1.80

all

-1.79

you

-1.73

-1.63

another

-1.56

-1.54

POSITIVE LOGITS

 into

2.19

derdag

1.75

).

1.73

骸

1.70

as

1.65

}$.

1.64

sweater

1.60

».

1.59

杙

1.58

The

1.53

Activations Density 0.031%