INDEX

Explanations

sorting, explanation, design, prior

The neuron fires on verbs naming tasks or actions (e.g. build, test, confirm, ask, explain).

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

to

-1.57

 Table

-1.34

 amateur

-1.33

 Figure

-1.31



-1.30

 που

-1.21

未完

-1.21

jaus

-1.19

 illustrious

-1.18

ukraine

-1.18

POSITIVE LOGITS

ogues

1.63

ayı

1.42

things

1.42

ed

1.40

in

1.38

 then

1.36

 things

1.34

byes

1.34

 vår

1.33

adays

1.30

Activations Density 0.002%