INDEX

Explanations

describes specific qualities

The neuron responds most strongly to unusually long words (i.e. tokens composed of many characters).

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

-1.89

),

-1.60

 waarvan

-1.36

 other

-1.34

 with

-1.34

To

-1.32

czeń

-1.30

</h2>

-1.29

........

-1.27

Some

-1.27

POSITIVE LOGITS

 erhi

1.42

 смогут

1.31

 zapato

1.30

 calzado

1.29

 nowy

1.28

 admiten

1.27

izaron

1.27

 rostros

1.27

を目指す

1.27

 suela

1.27

Activations Density 0.105%