INDEX

Explanations

introductions and next words

The neuron flags semantically important “content” tokens—especially domain-specific technical terms and other high-information words—throughout a document.

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

 like

-1.48

//!

-1.44

his

-1.39

 then

-1.37

 before

-1.27

 rozwią

-1.26

 السياس

-1.24

<tfoot>

-1.24

windig

-1.23

karet

-1.22

POSITIVE LOGITS

either

1.70

 podrían

1.59

малар

1.39

亟

1.36

 निम्नलिखित

1.27

~~

1.23

 tentar

1.23

especially

1.23

 figliu

1.22

usually

1.21

Activations Density 0.001%