INDEX

Explanations

relative pronouns

The neuron is picking out semantically rich “content” words (nouns, key verbs, adjectives, and pronouns) rather than common function words.

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

on

-1.41

 within

-1.29

 with

-1.20

 Süß

-1.18

 still

-1.17

at

-1.13

of

-1.12

Chimp

-1.12

 utama

-1.06

 sogar

-1.06

POSITIVE LOGITS

but

1.66

sterious

1.34

 другие

1.30



1.20

在一个

1.20

untura

1.20

 לְ

1.19

whom

1.18

 distinta

1.17

 viſ

1.17

Activations Density 0.074%