INDEX

Explanations

right and wrong

This neuron detects occurrences of the words “wrong” or “right,” i.e. explicit statements of something being correct or incorrect.

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

کاربرد

-1.58

our

-1.43

–

-1.41

镠

-1.41

ֹת

-1.40

Produktion

-1.38

给我们

-1.35

cN

-1.34

 aislados

-1.32

硖

-1.31

POSITIVE LOGITS

ങ്ങൾ

1.48

expliquer

1.48

>();

1.48

ጴ

1.46

 afir

1.45

it

1.44

 minori

1.43

も行

1.43

鏵

1.41

ائم

1.40

Activations Density 0.020%