INDEX

Explanations

describing a functional role

This neuron fires on the key nouns that label an entity’s functional role (e.g. “switch,” “filter,” “system,” “scaffolding,” “inhibitor,” “container”)—the words naming what something “acts as” or “serves as.”

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

 still

-1.20

屬於

-1.14

 that

-1.05

 lắm

-1.05

 acoged

-0.98

isShowing

-0.98

 wymaga

-0.94

 potrzebu

-0.94

 做法

-0.93

 gefragt

-0.93

POSITIVE LOGITS

for

1.32

ttt

1.18

tttt

1.13

 durante

1.05

hever

1.04

tothe

1.02

yPos

1.02

tral

1.02

εν

1.01

 during

1.00

Activations Density 0.045%