INDEX

Explanations

titles before names

The neuron fires on the abbreviation “Capt.” (i.e. the military rank “Captain”).

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

 Atención

-2.09

-1.77

itamente

-1.75

</table>

-1.72

乄

-1.69

acją

-1.66

ulitis

-1.65

鄠

-1.64

gonna

-1.60

ка

-1.59

POSITIVE LOGITS

 with

2.55

in

2.31

 which

2.17

for

2.14

who

1.93

 within

1.91

 based

1.91

 that

1.91

as

1.87

 then

1.85

Activations Density 0.008%