INDEX

Explanations

we.

The neuron strongly activates on the first‐person plural pronoun “we” (and closely related first‐person plural constructions).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

it

-1.72

-1.52

and

-1.43

 그리고

-1.33

 daughter

-1.29

 blah

-1.28

AND

-1.28

и

-1.26

stration

-1.25

 method

-1.23

POSITIVE LOGITS

 cœurs

1.44

Tetapi

1.42

“

1.38

 culmin

1.38

ᯅ

1.34

Από

1.33

Dette

1.31

hrten

1.31

Уже

1.31

 irres

1.30

Activations Density 0.031%