INDEX

Explanations

50 and quantities

The neuron fires on numeric tokens (numbers), especially multi-digit figures.

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

-2.81

 verwende

-2.63

The

-2.48

-2.47

 were

-2.47

“

-2.36

нет

-2.30

 adota

-2.28

the

-2.17

咴

-2.11

POSITIVE LOGITS

anf

2.69

‪.‬‬

2.58

 помогает

2.27

 стоковые

2.13

ündigung

2.11

 вони

2.05

.’

1.99

 フロア

1.99

 باسلام

1.97

1.96

Activations Density 0.043%