INDEX

Explanations

agreed to

The neuron fires on instances of the past-tense verb “agreed,” flagging mentions of formal agreement or consent.

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

⮜

-2.77

くちゃ

-2.58

 inim

-2.27

ндарт

-2.27

 να

-2.25

 hermanas

-2.25

 ворота

-2.22

“

-2.20

 DEPOSITORY

-2.19

POSITIVE LOGITS

3.27

With

2.63

2.47

What

2.36

These

2.34

AN

2.30

There

2.30

ان

2.27

,"

2.25

At

2.25

Activations Density 0.008%