INDEX

Explanations

with a

The neuron strongly activates on the preposition “with.”

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

to

-1.36

だが

-1.33

and

-1.28

 jedes

-1.27

がある

-1.23

 helft

-1.23

éget

-1.17

deauna

-1.16

varande

-1.15

 there

-1.14

POSITIVE LOGITS

3.63

 something

2.09

an

2.03

 either

1.89

 anúncio

1.70

 another

1.69

 enough

1.64

 capuz

1.60

 anjo

1.58

まさかの

1.58

Activations Density 0.116%